* [PATCH v6 0/8] prserv: add support for an "upstream" server
@ 2024-04-30 17:15 michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 1/8] prserv: declare "max_package_pr" client hook michael.opdenacker
` (8 more replies)
0 siblings, 9 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Thomas Petazzoni, Joshua Watt, Tim Orling
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
This makes it possible to customize an "upstream" distribution
by modifying local packages. If the "upstream" package bears
revision "x", the local one will have revision "x.y", this
having priority over the upstream one.
Multiple levels of upstream servers are supported, so "x.y.z" revisions
are possible too.
This version also supports multiple servers sharing the same database.
This also includes BitBake selftests to check the expected
functionality and detect future regressions.
Note that database import and export functions aren't tested yet.
---
Changes in V6:
- Harden the store_value() code to protect
it against storing the same value twice.
- Simplify the importone() hook, removing
its unnecessary "history" parameter.
- This fixes all errors in "oe-selftest -r prservice",
once this OE-core patch is merged:
https://lore.kernel.org/openembedded-core/20240429194656.655509-1-michael.opdenacker@bootlin.com/T/#u
- Make tests work in temporary directories.
Suggested by Joshua Watt
- Remove first version of readonly server tests
which depended on the output of a previous test.
One implementation of readonly server tests remains.
- Merge the upstream and read-only tests so that
the read-only test part can reuse the databases
and servers from the upstream one. Initially thought
that the setUp (and tearDown) hooks were called only
once for each test class, instead of for each test.
Thanks to Joshua Watt for making me understand this.
- Merged two more tests and simplified their naming
as, with one exception, the order of tests doesn't
matter any more.
- Avoid a possible race condition at database
creation time. Suggested by Joshua Watt.
Changes in V5:
- Change the way the database is accessed and modified.
As done in the Hash Server code, the database is not
accessed through the cursor API.
This makes it possible to have multiple servers accessing
the same database, typically one read-write server that
is fed by the build machine, and one read-only server that
is used to publish PR data to downstream users.
This switched sqlite3 to "WAL" journaling mode, and to
"OFF" synchronous mode.
- Fix revision ordering issues, now computing the min and max
values through dedicated Python functions, instead of letting
the database do this with just string comparison, which
was wrong in our case. Typically, "1.20" is greater for us
than "1.3".
- Add BitBake selftests, in particular to make sure revision
ordering is correct and the maximum and minimum values are
correctly computed.
- An issue remains: the BitBake selftests that test database
access through a read-only server while a read-write server
is still running, still fail, because the database changes
are apparently not stored to disk.
FAIL: test_3b_readonly (prserv.tests.PRUpstreamTests)
Changes in V4:
- Add BitBake selftests for the legacy and new PR server features
(database, client, server, support for upstream server, read-only mode, "history" and "no history" modes)
To run only these tests:
bitbake-selftest prserv.tests
- Pass the "history" mode through the client requests, instead
of storing it (globally) in the database name.
The PR database is now called "PRMAIN" instead of "PRMAIN_nohist".
This should cause a regression for builds which already
have a PR database.
- Fixes for "history" modes:
- Allow to store multiple PR values for the same checksum,
needed for the "no history" mode.
- Make the "history" mode return the minimum stored
PR value.
- Fixes and code reorganization for issues uncovered by the tests.
- Update the server version to "2.0.0"
Changes in V3:
- Revert the commit removing the so far unused "hist" mode, which
we wish to keep for binary reproducibility sake.
- Simplification of get_value() function to take
both "hist" and "nohist" modes with the same shared code.
- Add "history" parameter to the "getPR" request,
so that the client can ask for the mode of its choice.
This will also make it possible to implement tests
for both modes.
Note that more requests ("export", "import"...)
will also need a "history" parameter, in a future version,
after the first tests are implemented.
- Several bug fixes.
- Put all the new features at the tip of the branch,
to make the cleanup commits easier to merged.
Changes in V2:
- Add this new commit:
prserv: remove unused "hist" mode in the database backend
- Squash commit "prserv: fix read_only test" into
commit "prserv: simplify the PRServerClient() interface"
(Reported by Richard Purdie)
- Fix the code to support increasing "x.y.z" values, thus
supporting several levels of upstream servers.
- db.py: remove duplicate definition of find_max_value() function in db.py
- prserv.py: remove tabs before comments (Python didn't complain)
- db.py: now stores the revision ("value") as TEXT.
This way we can store "1.0" without having it transformed to "1"
when the default type was INTEGER.
- This allows to fix a regression when the first packages were created
with 'r0.1' instead of 'r0.0' initially.
- find_max_value: now returns None instead of '0' when no value is found
Before we couldn't tell the difference between a '0'
max value and the absence of such a value.
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Michael Opdenacker (8):
prserv: declare "max_package_pr" client hook
prserv: move code from __init__ to bitbake-prserv
prserv: add "upstream" server support
prserv: enable database sharing
prserv: avoid possible race condition in database code
prserv: store_value() improvements
prserv: import simplification
prserv: add bitbake selftests
bin/bitbake-prserv | 26 ++-
bin/bitbake-selftest | 2 +
lib/prserv/__init__.py | 97 ++++++++-
lib/prserv/client.py | 15 +-
lib/prserv/db.py | 452 ++++++++++++++++++-----------------------
lib/prserv/serv.py | 139 ++++++++++---
lib/prserv/tests.py | 386 +++++++++++++++++++++++++++++++++++
7 files changed, 816 insertions(+), 301 deletions(-)
create mode 100644 bitbake/lib/prserv/tests.py
--
2.34.1
^ permalink raw reply [flat|nested] 11+ messages in thread
* [PATCH v6 1/8] prserv: declare "max_package_pr" client hook
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 2/8] prserv: move code from __init__ to bitbake-prserv michael.opdenacker
` (7 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
Add missing declaration for the max_package_pr client hook
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
lib/prserv/client.py | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/lib/prserv/client.py b/lib/prserv/client.py
index 8471ee3046..99fc4e0f7f 100644
--- a/lib/prserv/client.py
+++ b/lib/prserv/client.py
@@ -65,7 +65,7 @@ class PRAsyncClient(bb.asyncrpc.AsyncClient):
class PRClient(bb.asyncrpc.Client):
def __init__(self):
super().__init__()
- self._add_methods("getPR", "test_pr", "test_package", "importone", "export", "is_readonly")
+ self._add_methods("getPR", "test_pr", "test_package", "max_package_pr", "importone", "export", "is_readonly")
def _get_async_client(self):
return PRAsyncClient()
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 2/8] prserv: move code from __init__ to bitbake-prserv
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 1/8] prserv: declare "max_package_pr" client hook michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 3/8] prserv: add "upstream" server support michael.opdenacker
` (6 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
This script was the only user of this code.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
bin/bitbake-prserv | 9 ++++++++-
lib/prserv/__init__.py | 13 -------------
2 files changed, 8 insertions(+), 14 deletions(-)
diff --git a/bin/bitbake-prserv b/bin/bitbake-prserv
index ad0a069401..920663a1d8 100755
--- a/bin/bitbake-prserv
+++ b/bin/bitbake-prserv
@@ -21,6 +21,13 @@ VERSION = "1.1.0"
PRHOST_DEFAULT="0.0.0.0"
PRPORT_DEFAULT=8585
+def init_logger(logfile, loglevel):
+ numeric_level = getattr(logging, loglevel.upper(), None)
+ if not isinstance(numeric_level, int):
+ raise ValueError("Invalid log level: %s" % loglevel)
+ FORMAT = "%(asctime)-15s %(message)s"
+ logging.basicConfig(level=numeric_level, filename=logfile, format=FORMAT)
+
def main():
parser = argparse.ArgumentParser(
description="BitBake PR Server. Version=%s" % VERSION,
@@ -72,7 +79,7 @@ def main():
)
args = parser.parse_args()
- prserv.init_logger(os.path.abspath(args.log), args.loglevel)
+ init_logger(os.path.abspath(args.log), args.loglevel)
if args.start:
ret=prserv.serv.start_daemon(args.file, args.host, args.port, os.path.abspath(args.log), args.read_only)
diff --git a/lib/prserv/__init__.py b/lib/prserv/__init__.py
index 0e0aa34d0e..94658b815d 100644
--- a/lib/prserv/__init__.py
+++ b/lib/prserv/__init__.py
@@ -5,16 +5,3 @@
#
__version__ = "1.0.0"
-
-import os, time
-import sys, logging
-
-def init_logger(logfile, loglevel):
- numeric_level = getattr(logging, loglevel.upper(), None)
- if not isinstance(numeric_level, int):
- raise ValueError("Invalid log level: %s" % loglevel)
- FORMAT = "%(asctime)-15s %(message)s"
- logging.basicConfig(level=numeric_level, filename=logfile, format=FORMAT)
-
-class NotFoundError(Exception):
- pass
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 3/8] prserv: add "upstream" server support
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 1/8] prserv: declare "max_package_pr" client hook michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 2/8] prserv: move code from __init__ to bitbake-prserv michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 4/8] prserv: enable database sharing michael.opdenacker
` (5 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
Introduce a PRSERVER_UPSTREAM variable that makes the
local PR server connect to an "upstream" one.
This makes it possible to implement local fixes to an
upstream package (revision "x", in a way that gives the local
update priority (revision "x.y").
Update the calculation of the new revisions to support the
case when prior revisions are not integers, but have
an "x.y..." format."
Set the comments in the handle_get_pr() function in serv.py
for details about the calculation of the local revision.
This is done by going on supporting the "history" mode that
wasn't used so far (revisions can return to a previous historical value),
in addition to the default "no history" mode (revisions can never decrease).
Rather than storing the history mode in the database table
itself (i.e. "PRMAIN_hist" and "PRMAIN_nohist"), the history mode
is now passed through the client requests. As a consequence, the
table name is now "PRMAIN", which is incompatible with what
was generated before, but avoids confusion if we kept the "PRMAIN_nohist"
name for both "history" and "no history" modes.
Update the server version to "2.0.0".
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
bin/bitbake-prserv | 17 +++-
lib/prserv/__init__.py | 90 +++++++++++++++++-
lib/prserv/client.py | 17 ++--
lib/prserv/db.py | 205 ++++++++++++++++++++++-------------------
lib/prserv/serv.py | 134 +++++++++++++++++++++++----
5 files changed, 338 insertions(+), 125 deletions(-)
diff --git a/bin/bitbake-prserv b/bin/bitbake-prserv
index 920663a1d8..580e021fda 100755
--- a/bin/bitbake-prserv
+++ b/bin/bitbake-prserv
@@ -16,7 +16,7 @@ sys.path.insert(0, os.path.join(os.path.dirname(os.path.dirname(__file__)), "lib
import prserv
import prserv.serv
-VERSION = "1.1.0"
+VERSION = "2.0.0"
PRHOST_DEFAULT="0.0.0.0"
PRPORT_DEFAULT=8585
@@ -77,12 +77,25 @@ def main():
action="store_true",
help="open database in read-only mode",
)
+ parser.add_argument(
+ "-u",
+ "--upstream",
+ default=os.environ.get("PRSERVER_UPSTREAM", None),
+ help="Upstream PR service (host:port)",
+ )
args = parser.parse_args()
init_logger(os.path.abspath(args.log), args.loglevel)
if args.start:
- ret=prserv.serv.start_daemon(args.file, args.host, args.port, os.path.abspath(args.log), args.read_only)
+ ret=prserv.serv.start_daemon(
+ args.file,
+ args.host,
+ args.port,
+ os.path.abspath(args.log),
+ args.read_only,
+ args.upstream
+ )
elif args.stop:
ret=prserv.serv.stop_daemon(args.host, args.port)
else:
diff --git a/lib/prserv/__init__.py b/lib/prserv/__init__.py
index 94658b815d..a817b03c1e 100644
--- a/lib/prserv/__init__.py
+++ b/lib/prserv/__init__.py
@@ -4,4 +4,92 @@
# SPDX-License-Identifier: GPL-2.0-only
#
-__version__ = "1.0.0"
+
+__version__ = "2.0.0"
+
+import logging
+logger = logging.getLogger("BitBake.PRserv")
+
+from bb.asyncrpc.client import parse_address, ADDR_TYPE_UNIX, ADDR_TYPE_WS
+
+def create_server(addr, dbpath, upstream=None, read_only=False):
+ from . import serv
+
+ s = serv.PRServer(dbpath, upstream=upstream, read_only=read_only)
+ host, port = addr.split(":")
+ s.start_tcp_server(host, int(port))
+
+ return s
+
+def increase_revision(ver):
+ """Take a revision string such as "1" or "1.2.3" or even a number and increase its last number
+ This fails if the last number is not an integer"""
+
+ fields=str(ver).split('.')
+ last = fields[-1]
+
+ try:
+ val = int(last)
+ except Exception as e:
+ logger.critical("Unable to increase revision value %s: %s" % (ver, e))
+ raise e
+
+ return ".".join(fields[0:-1] + list(str(val + 1)))
+
+def _revision_greater_or_equal(rev1, rev2):
+ """Compares x.y.z revision numbers, using integer comparison
+ Returns True if rev1 is greater or equal to rev2"""
+
+ fields1 = rev1.split(".")
+ fields2 = rev2.split(".")
+ l1 = len(fields1)
+ l2 = len(fields2)
+
+ for i in range(l1):
+ val1 = int(fields1[i])
+ if i < l2:
+ val2 = int(fields2[i])
+ if val2 < val1:
+ return True
+ elif val2 > val1:
+ return False
+ else:
+ return True
+ return True
+
+def revision_smaller(rev1, rev2):
+ """Compares x.y.z revision numbers, using integer comparison
+ Returns True if rev1 is strictly smaller than rev2"""
+ return not(_revision_greater_or_equal(rev1, rev2))
+
+def revision_greater(rev1, rev2):
+ """Compares x.y.z revision numbers, using integer comparison
+ Returns True if rev1 is strictly greater than rev2"""
+ return _revision_greater_or_equal(rev1, rev2) and (rev1 != rev2)
+
+def create_client(addr):
+ from . import client
+
+ c = client.PRClient()
+
+ try:
+ (typ, a) = parse_address(addr)
+ c.connect_tcp(*a)
+ return c
+ except Exception as e:
+ c.close()
+ raise e
+
+async def create_async_client(addr):
+ from . import client
+
+ c = client.PRAsyncClient()
+
+ try:
+ (typ, a) = parse_address(addr)
+ await c.connect_tcp(*a)
+ return c
+
+ except Exception as e:
+ await c.close()
+ raise e
diff --git a/lib/prserv/client.py b/lib/prserv/client.py
index 99fc4e0f7f..565c6f3872 100644
--- a/lib/prserv/client.py
+++ b/lib/prserv/client.py
@@ -6,6 +6,7 @@
import logging
import bb.asyncrpc
+from . import create_async_client
logger = logging.getLogger("BitBake.PRserv")
@@ -13,16 +14,16 @@ class PRAsyncClient(bb.asyncrpc.AsyncClient):
def __init__(self):
super().__init__("PRSERVICE", "1.0", logger)
- async def getPR(self, version, pkgarch, checksum):
+ async def getPR(self, version, pkgarch, checksum, history=False):
response = await self.invoke(
- {"get-pr": {"version": version, "pkgarch": pkgarch, "checksum": checksum}}
+ {"get-pr": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "history": history}}
)
if response:
return response["value"]
- async def test_pr(self, version, pkgarch, checksum):
+ async def test_pr(self, version, pkgarch, checksum, history=False):
response = await self.invoke(
- {"test-pr": {"version": version, "pkgarch": pkgarch, "checksum": checksum}}
+ {"test-pr": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "history": history}}
)
if response:
return response["value"]
@@ -41,16 +42,16 @@ class PRAsyncClient(bb.asyncrpc.AsyncClient):
if response:
return response["value"]
- async def importone(self, version, pkgarch, checksum, value):
+ async def importone(self, version, pkgarch, checksum, value, history=False):
response = await self.invoke(
- {"import-one": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "value": value}}
+ {"import-one": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "value": value, "history": history}}
)
if response:
return response["value"]
- async def export(self, version, pkgarch, checksum, colinfo):
+ async def export(self, version, pkgarch, checksum, colinfo, history=False):
response = await self.invoke(
- {"export": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "colinfo": colinfo}}
+ {"export": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "colinfo": colinfo, "history": history}}
)
if response:
return (response["metainfo"], response["datainfo"])
diff --git a/lib/prserv/db.py b/lib/prserv/db.py
index eb41508198..b2520f3158 100644
--- a/lib/prserv/db.py
+++ b/lib/prserv/db.py
@@ -10,6 +10,8 @@ import errno
import prserv
import time
+from . import increase_revision, revision_greater, revision_smaller
+
try:
import sqlite3
except ImportError:
@@ -32,15 +34,11 @@ if sqlversion[0] < 3 or (sqlversion[0] == 3 and sqlversion[1] < 3):
#
class PRTable(object):
- def __init__(self, conn, table, nohist, read_only):
+ def __init__(self, conn, table, read_only):
self.conn = conn
- self.nohist = nohist
self.read_only = read_only
self.dirty = False
- if nohist:
- self.table = "%s_nohist" % table
- else:
- self.table = "%s_hist" % table
+ self.table = table
if self.read_only:
table_exists = self._execute(
@@ -53,8 +51,8 @@ class PRTable(object):
(version TEXT NOT NULL, \
pkgarch TEXT NOT NULL, \
checksum TEXT NOT NULL, \
- value INTEGER, \
- PRIMARY KEY (version, pkgarch, checksum));" % self.table)
+ value TEXT, \
+ PRIMARY KEY (version, pkgarch, checksum, value));" % self.table)
def _execute(self, *query):
"""Execute a query, waiting to acquire a lock if necessary"""
@@ -68,6 +66,28 @@ class PRTable(object):
continue
raise exc
+ def _extremum_value(self, rows, is_max):
+ value = None
+
+ for row in rows:
+ current_value = row[0]
+ if value is None:
+ value = current_value
+ else:
+ if is_max:
+ is_new_extremum = revision_greater(current_value, value)
+ else:
+ is_new_extremum = revision_smaller(current_value, value)
+ if is_new_extremum:
+ value = current_value
+ return value
+
+ def _max_value(self, rows):
+ return self._extremum_value(rows, True)
+
+ def _min_value(self, rows):
+ return self._extremum_value(rows, False)
+
def sync(self):
if not self.read_only:
self.conn.commit()
@@ -102,101 +122,93 @@ class PRTable(object):
else:
return False
- def find_value(self, version, pkgarch, checksum):
+
+ def find_package_max_value(self, version, pkgarch):
+ """Returns the greatest value for (version, pkgarch), or None if not found. Doesn't create a new value"""
+
+ data = self._execute("SELECT value FROM %s where version=? AND pkgarch=?;" % (self.table),
+ (version, pkgarch))
+ rows = data.fetchall()
+ value = self._max_value(rows)
+ return value
+
+ def find_value(self, version, pkgarch, checksum, history=False):
"""Returns the value for the specified checksum if found or None otherwise."""
- data=self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
- (version, pkgarch, checksum))
- row=data.fetchone()
- if row is not None:
- return row[0]
+ if history:
+ return self.find_min_value(version, pkgarch, checksum)
else:
- return None
+ return self.find_max_value(version, pkgarch, checksum)
- def find_max_value(self, version, pkgarch):
- """Returns the greatest value for (version, pkgarch), or None if not found. Doesn't create a new value"""
- data = self._execute("SELECT max(value) FROM %s where version=? AND pkgarch=?;" % (self.table),
+ def _find_extremum_value(self, version, pkgarch, checksum, is_max):
+ """Returns the maximum (if is_max is True) or minimum (if is_max is False) value
+ for (version, pkgarch, checksum), or None if not found. Doesn't create a new value"""
+
+ data = self._execute("SELECT value FROM %s where version=? AND pkgarch=? AND checksum=?;" % (self.table),
+ (version, pkgarch, checksum))
+ rows = data.fetchall()
+ return self._extremum_value(rows, is_max)
+
+ def find_max_value(self, version, pkgarch, checksum):
+ return self._find_extremum_value(version, pkgarch, checksum, True)
+
+ def find_min_value(self, version, pkgarch, checksum):
+ return self._find_extremum_value(version, pkgarch, checksum, False)
+
+ def find_new_subvalue(self, version, pkgarch, base):
+ """Take and increase the greatest "<base>.y" value for (version, pkgarch), or return "<base>.0" if not found.
+ This doesn't store a new value."""
+
+ data = self._execute("SELECT value FROM %s where version=? AND pkgarch=? AND value LIKE '%s.%%';" % (self.table, base),
(version, pkgarch))
- row = data.fetchone()
- if row is not None:
- return row[0]
- else:
- return None
+ rows = data.fetchall()
+ value = self._max_value(rows)
- def _get_value_hist(self, version, pkgarch, checksum):
- data=self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
- (version, pkgarch, checksum))
- row=data.fetchone()
- if row is not None:
- return row[0]
+ if value is not None:
+ return increase_revision(value)
else:
- #no value found, try to insert
- if self.read_only:
- data = self._execute("SELECT ifnull(max(value)+1, 0) FROM %s where version=? AND pkgarch=?;" % (self.table),
- (version, pkgarch))
- row = data.fetchone()
- if row is not None:
- return row[0]
- else:
- return 0
+ return base + ".0"
- try:
- self._execute("INSERT INTO %s VALUES (?, ?, ?, (select ifnull(max(value)+1, 0) from %s where version=? AND pkgarch=?));"
- % (self.table, self.table),
- (version, pkgarch, checksum, version, pkgarch))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
+ def store_value(self, version, pkgarch, checksum, value):
+ """Store new value in the database"""
- self.dirty = True
+ try:
+ self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
+ (version, pkgarch, checksum, value))
+ except sqlite3.IntegrityError as exc:
+ logger.error(str(exc))
- data=self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
- (version, pkgarch, checksum))
- row=data.fetchone()
- if row is not None:
- return row[0]
- else:
- raise prserv.NotFoundError
+ self.dirty = True
- def _get_value_no_hist(self, version, pkgarch, checksum):
- data=self._execute("SELECT value FROM %s \
- WHERE version=? AND pkgarch=? AND checksum=? AND \
- value >= (select max(value) from %s where version=? AND pkgarch=?);"
- % (self.table, self.table),
- (version, pkgarch, checksum, version, pkgarch))
- row=data.fetchone()
- if row is not None:
- return row[0]
- else:
- #no value found, try to insert
- if self.read_only:
- data = self._execute("SELECT ifnull(max(value)+1, 0) FROM %s where version=? AND pkgarch=?;" % (self.table),
- (version, pkgarch))
- return data.fetchone()[0]
+ def _get_value(self, version, pkgarch, checksum, history):
- try:
- self._execute("INSERT OR REPLACE INTO %s VALUES (?, ?, ?, (select ifnull(max(value)+1, 0) from %s where version=? AND pkgarch=?));"
- % (self.table, self.table),
- (version, pkgarch, checksum, version, pkgarch))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
- self.conn.rollback()
+ max_value = self.find_package_max_value(version, pkgarch)
- self.dirty = True
+ if max_value is None:
+ # version, pkgarch completely unknown. Return initial value.
+ return "0"
- data=self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
- (version, pkgarch, checksum))
- row=data.fetchone()
- if row is not None:
- return row[0]
- else:
- raise prserv.NotFoundError
+ value = self.find_value(version, pkgarch, checksum, history)
- def get_value(self, version, pkgarch, checksum):
- if self.nohist:
- return self._get_value_no_hist(version, pkgarch, checksum)
+ if value is None:
+ # version, pkgarch found but not checksum. Create a new value from the maximum one
+ return increase_revision(max_value)
+
+ if history:
+ return value
+
+ # "no history" mode - If the value is not the maximum value for the package, need to increase it.
+ if max_value > value:
+ return increase_revision(max_value)
else:
- return self._get_value_hist(version, pkgarch, checksum)
+ return value
+
+ def get_value(self, version, pkgarch, checksum, history):
+ value = self._get_value(version, pkgarch, checksum, history)
+ if not self.read_only:
+ self.store_value(version, pkgarch, checksum, value)
+ return value
def _import_hist(self, version, pkgarch, checksum, value):
if self.read_only:
@@ -252,13 +264,13 @@ class PRTable(object):
else:
return None
- def importone(self, version, pkgarch, checksum, value):
- if self.nohist:
- return self._import_no_hist(version, pkgarch, checksum, value)
- else:
+ def importone(self, version, pkgarch, checksum, value, history=False):
+ if history:
return self._import_hist(version, pkgarch, checksum, value)
+ else:
+ return self._import_no_hist(version, pkgarch, checksum, value)
- def export(self, version, pkgarch, checksum, colinfo):
+ def export(self, version, pkgarch, checksum, colinfo, history=False):
metainfo = {}
#column info
if colinfo:
@@ -278,12 +290,12 @@ class PRTable(object):
#data info
datainfo = []
- if self.nohist:
+ if history:
+ sqlstmt = "SELECT * FROM %s as T1 WHERE 1=1 " % self.table
+ else:
sqlstmt = "SELECT T1.version, T1.pkgarch, T1.checksum, T1.value FROM %s as T1, \
(SELECT version, pkgarch, max(value) as maxvalue FROM %s GROUP BY version, pkgarch) as T2 \
WHERE T1.version=T2.version AND T1.pkgarch=T2.pkgarch AND T1.value=T2.maxvalue " % (self.table, self.table)
- else:
- sqlstmt = "SELECT * FROM %s as T1 WHERE 1=1 " % self.table
sqlarg = []
where = ""
if version:
@@ -322,9 +334,8 @@ class PRTable(object):
class PRData(object):
"""Object representing the PR database"""
- def __init__(self, filename, nohist=True, read_only=False):
+ def __init__(self, filename, read_only=False):
self.filename=os.path.abspath(filename)
- self.nohist=nohist
self.read_only = read_only
#build directory hierarchy
try:
@@ -351,7 +362,7 @@ class PRData(object):
if tblname in self._tables:
return self._tables[tblname]
else:
- tableobj = self._tables[tblname] = PRTable(self.connection, tblname, self.nohist, self.read_only)
+ tableobj = self._tables[tblname] = PRTable(self.connection, tblname, self.read_only)
return tableobj
def __delitem__(self, tblname):
diff --git a/lib/prserv/serv.py b/lib/prserv/serv.py
index dc4be5b620..05573d06cc 100644
--- a/lib/prserv/serv.py
+++ b/lib/prserv/serv.py
@@ -12,6 +12,7 @@ import sqlite3
import prserv
import prserv.db
import errno
+from . import create_async_client, revision_smaller, increase_revision
import bb.asyncrpc
logger = logging.getLogger("BitBake.PRserv")
@@ -51,8 +52,9 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
version = request["version"]
pkgarch = request["pkgarch"]
checksum = request["checksum"]
+ history = request["history"]
- value = self.server.table.find_value(version, pkgarch, checksum)
+ value = self.server.table.find_value(version, pkgarch, checksum, history)
return {"value": value}
async def handle_test_package(self, request):
@@ -68,22 +70,110 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
version = request["version"]
pkgarch = request["pkgarch"]
- value = self.server.table.find_max_value(version, pkgarch)
+ value = self.server.table.find_package_max_value(version, pkgarch)
return {"value": value}
async def handle_get_pr(self, request):
version = request["version"]
pkgarch = request["pkgarch"]
checksum = request["checksum"]
+ history = request["history"]
- response = None
- try:
- value = self.server.table.get_value(version, pkgarch, checksum)
- response = {"value": value}
- except prserv.NotFoundError:
- self.logger.error("failure storing value in database for (%s, %s)",version, checksum)
+ if self.upstream_client is None:
+ value = self.server.table.get_value(version, pkgarch, checksum, history)
+ return {"value": value}
- return response
+ # We have an upstream server.
+ # Check whether the local server already knows the requested configuration.
+ # If the configuration is a new one, the generated value we will add will
+ # depend on what's on the upstream server. That's why we're calling find_value()
+ # instead of get_value() directly.
+
+ value = self.server.table.find_value(version, pkgarch, checksum, history)
+ upstream_max = await self.upstream_client.max_package_pr(version, pkgarch)
+
+ if value is not None:
+
+ # The configuration is already known locally.
+
+ if history:
+ value = self.server.table.get_value(version, pkgarch, checksum, history)
+ else:
+ existing_value = value
+ # In "no history", we need to make sure the value doesn't decrease
+ # and is at least greater than the maximum upstream value
+ # and the maximum local value
+
+ local_max = self.server.table.find_package_max_value(version, pkgarch)
+ if revision_smaller(value, local_max):
+ value = increase_revision(local_max)
+
+ if revision_smaller(value, upstream_max):
+ # Ask upstream whether it knows the checksum
+ upstream_value = await self.upstream_client.test_pr(version, pkgarch, checksum)
+ if upstream_value is None:
+ # Upstream doesn't have our checksum, let create a new one
+ value = upstream_max + ".0"
+ else:
+ # Fine to take the same value as upstream
+ value = upstream_max
+
+ if not value == existing_value and not self.server.read_only:
+ self.server.table.store_value(version, pkgarch, checksum, value)
+
+ return {"value": value}
+
+ # The configuration is a new one for the local server
+ # Let's ask the upstream server whether it knows it
+
+ known_upstream = await self.upstream_client.test_package(version, pkgarch)
+
+ if not known_upstream:
+
+ # The package is not known upstream, must be a local-only package
+ # Let's compute the PR number using the local-only method
+
+ value = self.server.table.get_value(version, pkgarch, checksum, history)
+ return {"value": value}
+
+ # The package is known upstream, let's ask the upstream server
+ # whether it knows our new output hash
+
+ value = await self.upstream_client.test_pr(version, pkgarch, checksum)
+
+ if value is not None:
+
+ # Upstream knows this output hash, let's store it and use it too.
+
+ if not self.server.read_only:
+ self.server.table.store_value(version, pkgarch, checksum, value)
+ # If the local server is read only, won't be able to store the new
+ # value in the database and will have to keep asking the upstream server
+ return {"value": value}
+
+ # The output hash doesn't exist upstream, get the most recent number from upstream (x)
+ # Then, we want to have a new PR value for the local server: x.y
+
+ upstream_max = await self.upstream_client.max_package_pr(version, pkgarch)
+ # Here we know that the package is known upstream, so upstream_max can't be None
+ subvalue = self.server.table.find_new_subvalue(version, pkgarch, upstream_max)
+
+ if not self.server.read_only:
+ self.server.table.store_value(version, pkgarch, checksum, subvalue)
+
+ return {"value": subvalue}
+
+ async def process_requests(self):
+ if self.server.upstream is not None:
+ self.upstream_client = await create_async_client(self.server.upstream)
+ else:
+ self.upstream_client = None
+
+ try:
+ await super().process_requests()
+ finally:
+ if self.upstream_client is not None:
+ await self.upstream_client.close()
async def handle_import_one(self, request):
response = None
@@ -92,8 +182,9 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
pkgarch = request["pkgarch"]
checksum = request["checksum"]
value = request["value"]
+ history = request["history"]
- value = self.server.table.importone(version, pkgarch, checksum, value)
+ value = self.server.table.importone(version, pkgarch, checksum, value, history)
if value is not None:
response = {"value": value}
@@ -104,9 +195,10 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
pkgarch = request["pkgarch"]
checksum = request["checksum"]
colinfo = request["colinfo"]
+ history = request["history"]
try:
- (metainfo, datainfo) = self.server.table.export(version, pkgarch, checksum, colinfo)
+ (metainfo, datainfo) = self.server.table.export(version, pkgarch, checksum, colinfo, history)
except sqlite3.Error as exc:
self.logger.error(str(exc))
metainfo = datainfo = None
@@ -117,11 +209,12 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
return {"readonly": self.server.read_only}
class PRServer(bb.asyncrpc.AsyncServer):
- def __init__(self, dbfile, read_only=False):
+ def __init__(self, dbfile, read_only=False, upstream=None):
super().__init__(logger)
self.dbfile = dbfile
self.table = None
self.read_only = read_only
+ self.upstream = upstream
def accept_client(self, socket):
return PRServerClient(socket, self)
@@ -134,6 +227,9 @@ class PRServer(bb.asyncrpc.AsyncServer):
self.logger.info("Started PRServer with DBfile: %s, Address: %s, PID: %s" %
(self.dbfile, self.address, str(os.getpid())))
+ if self.upstream is not None:
+ self.logger.info("And upstream PRServer: %s " % (self.upstream))
+
return tasks
async def stop(self):
@@ -147,14 +243,15 @@ class PRServer(bb.asyncrpc.AsyncServer):
self.table.sync()
class PRServSingleton(object):
- def __init__(self, dbfile, logfile, host, port):
+ def __init__(self, dbfile, logfile, host, port, upstream):
self.dbfile = dbfile
self.logfile = logfile
self.host = host
self.port = port
+ self.upstream = upstream
def start(self):
- self.prserv = PRServer(self.dbfile)
+ self.prserv = PRServer(self.dbfile, upstream=self.upstream)
self.prserv.start_tcp_server(socket.gethostbyname(self.host), self.port)
self.process = self.prserv.serve_as_process(log_level=logging.WARNING)
@@ -233,7 +330,7 @@ def run_as_daemon(func, pidfile, logfile):
os.remove(pidfile)
os._exit(0)
-def start_daemon(dbfile, host, port, logfile, read_only=False):
+def start_daemon(dbfile, host, port, logfile, read_only=False, upstream=None):
ip = socket.gethostbyname(host)
pidfile = PIDPREFIX % (ip, port)
try:
@@ -249,7 +346,7 @@ def start_daemon(dbfile, host, port, logfile, read_only=False):
dbfile = os.path.abspath(dbfile)
def daemon_main():
- server = PRServer(dbfile, read_only=read_only)
+ server = PRServer(dbfile, read_only=read_only, upstream=upstream)
server.start_tcp_server(ip, port)
server.serve_forever()
@@ -336,6 +433,9 @@ def auto_start(d):
host = host_params[0].strip().lower()
port = int(host_params[1])
+
+ upstream = d.getVar("PRSERV_UPSTREAM") or None
+
if is_local_special(host, port):
import bb.utils
cachedir = (d.getVar("PERSISTENT_DIR") or d.getVar("CACHE"))
@@ -350,7 +450,7 @@ def auto_start(d):
auto_shutdown()
if not singleton:
bb.utils.mkdirhier(cachedir)
- singleton = PRServSingleton(os.path.abspath(dbfile), os.path.abspath(logfile), host, port)
+ singleton = PRServSingleton(os.path.abspath(dbfile), os.path.abspath(logfile), host, port, upstream)
singleton.start()
if singleton:
host = singleton.host
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 4/8] prserv: enable database sharing
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
` (2 preceding siblings ...)
2024-04-30 17:15 ` [PATCH v6 3/8] prserv: add "upstream" server support michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 20:29 ` [bitbake-devel] " Jan-Simon Moeller
2024-04-30 17:15 ` [PATCH v6 5/8] prserv: avoid possible race condition in database code michael.opdenacker
` (4 subsequent siblings)
8 siblings, 1 reply; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
sqlite3 can allow multiple processes to access the database
simultaneously, but it must be opened correctly. The key change is that
the database is no longer opened in "exclusive" mode (defaulting to
shared mode). In addition, the journal is set to "WAL" mode, as this is
the most efficient for dealing with simultaneous access between
different processes. In order to keep the database performance,
synchronous mode is set to "off". The WAL journal will protect against
incomplete transactions in any given client, however the database will
not be protected against unexpected power loss from the OS (which is a
fine trade off for performance, and also the same as the previous
implementation).
The use of a database cursor enabled to remove the _execute() wrapper.
The cursor automatically makes sure that the query happens in an atomic
transaction and commits when finished.
This also removes the need for a "dirty" flag for the database and
for explicit database syncing, which simplifies the code.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
lib/prserv/db.py | 322 +++++++++++++++++++++------------------------
lib/prserv/serv.py | 8 --
2 files changed, 151 insertions(+), 179 deletions(-)
diff --git a/lib/prserv/db.py b/lib/prserv/db.py
index b2520f3158..f430586d73 100644
--- a/lib/prserv/db.py
+++ b/lib/prserv/db.py
@@ -8,21 +8,13 @@ import logging
import os.path
import errno
import prserv
-import time
+import sqlite3
+from contextlib import closing
from . import increase_revision, revision_greater, revision_smaller
-try:
- import sqlite3
-except ImportError:
- from pysqlite2 import dbapi2 as sqlite3
-
logger = logging.getLogger("BitBake.PRserv")
-sqlversion = sqlite3.sqlite_version_info
-if sqlversion[0] < 3 or (sqlversion[0] == 3 and sqlversion[1] < 3):
- raise Exception("sqlite3 version 3.3.0 or later is required.")
-
#
# "No History" mode - for a given query tuple (version, pkgarch, checksum),
# the returned value will be the largest among all the values of the same
@@ -31,40 +23,28 @@ if sqlversion[0] < 3 or (sqlversion[0] == 3 and sqlversion[1] < 3):
# "History" mode - Return a new higher value for previously unseen query
# tuple (version, pkgarch, checksum), otherwise return historical value.
# Value can decrement if returning to a previous build.
-#
class PRTable(object):
def __init__(self, conn, table, read_only):
self.conn = conn
self.read_only = read_only
- self.dirty = False
self.table = table
- if self.read_only:
- table_exists = self._execute(
- "SELECT count(*) FROM sqlite_master \
- WHERE type='table' AND name='%s'" % (self.table))
- if not table_exists:
- raise prserv.NotFoundError
- else:
- self._execute("CREATE TABLE IF NOT EXISTS %s \
- (version TEXT NOT NULL, \
- pkgarch TEXT NOT NULL, \
- checksum TEXT NOT NULL, \
- value TEXT, \
- PRIMARY KEY (version, pkgarch, checksum, value));" % self.table)
-
- def _execute(self, *query):
- """Execute a query, waiting to acquire a lock if necessary"""
- start = time.time()
- end = start + 20
- while True:
- try:
- return self.conn.execute(*query)
- except sqlite3.OperationalError as exc:
- if "is locked" in str(exc) and end > time.time():
- continue
- raise exc
+ with closing(self.conn.cursor()) as cursor:
+ if self.read_only:
+ table_exists = cursor.execute(
+ "SELECT count(*) FROM sqlite_master \
+ WHERE type='table' AND name='%s'" % (self.table))
+ if not table_exists:
+ raise prserv.NotFoundError
+ else:
+ cursor.execute("CREATE TABLE IF NOT EXISTS %s \
+ (version TEXT NOT NULL, \
+ pkgarch TEXT NOT NULL, \
+ checksum TEXT NOT NULL, \
+ value TEXT, \
+ PRIMARY KEY (version, pkgarch, checksum, value));" % self.table)
+ self.conn.commit()
def _extremum_value(self, rows, is_max):
value = None
@@ -88,49 +68,42 @@ class PRTable(object):
def _min_value(self, rows):
return self._extremum_value(rows, False)
- def sync(self):
- if not self.read_only:
- self.conn.commit()
- self._execute("BEGIN EXCLUSIVE TRANSACTION")
-
- def sync_if_dirty(self):
- if self.dirty:
- self.sync()
- self.dirty = False
-
def test_package(self, version, pkgarch):
"""Returns whether the specified package version is found in the database for the specified architecture"""
# Just returns the value if found or None otherwise
- data=self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=?;" % self.table,
- (version, pkgarch))
- row=data.fetchone()
- if row is not None:
- return True
- else:
- return False
+ with closing(self.conn.cursor()) as cursor:
+ data=cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=?;" % self.table,
+ (version, pkgarch))
+ row=data.fetchone()
+ if row is not None:
+ return True
+ else:
+ return False
def test_value(self, version, pkgarch, value):
"""Returns whether the specified value is found in the database for the specified package and architecture"""
# Just returns the value if found or None otherwise
- data=self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? and value=?;" % self.table,
- (version, pkgarch, value))
- row=data.fetchone()
- if row is not None:
- return True
- else:
- return False
+ with closing(self.conn.cursor()) as cursor:
+ data=cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? and value=?;" % self.table,
+ (version, pkgarch, value))
+ row=data.fetchone()
+ if row is not None:
+ return True
+ else:
+ return False
def find_package_max_value(self, version, pkgarch):
"""Returns the greatest value for (version, pkgarch), or None if not found. Doesn't create a new value"""
- data = self._execute("SELECT value FROM %s where version=? AND pkgarch=?;" % (self.table),
- (version, pkgarch))
- rows = data.fetchall()
- value = self._max_value(rows)
- return value
+ with closing(self.conn.cursor()) as cursor:
+ data = cursor.execute("SELECT value FROM %s where version=? AND pkgarch=?;" % (self.table),
+ (version, pkgarch))
+ rows = data.fetchall()
+ value = self._max_value(rows)
+ return value
def find_value(self, version, pkgarch, checksum, history=False):
"""Returns the value for the specified checksum if found or None otherwise."""
@@ -145,10 +118,11 @@ class PRTable(object):
"""Returns the maximum (if is_max is True) or minimum (if is_max is False) value
for (version, pkgarch, checksum), or None if not found. Doesn't create a new value"""
- data = self._execute("SELECT value FROM %s where version=? AND pkgarch=? AND checksum=?;" % (self.table),
- (version, pkgarch, checksum))
- rows = data.fetchall()
- return self._extremum_value(rows, is_max)
+ with closing(self.conn.cursor()) as cursor:
+ data = cursor.execute("SELECT value FROM %s where version=? AND pkgarch=? AND checksum=?;" % (self.table),
+ (version, pkgarch, checksum))
+ rows = data.fetchall()
+ return self._extremum_value(rows, is_max)
def find_max_value(self, version, pkgarch, checksum):
return self._find_extremum_value(version, pkgarch, checksum, True)
@@ -160,26 +134,27 @@ class PRTable(object):
"""Take and increase the greatest "<base>.y" value for (version, pkgarch), or return "<base>.0" if not found.
This doesn't store a new value."""
- data = self._execute("SELECT value FROM %s where version=? AND pkgarch=? AND value LIKE '%s.%%';" % (self.table, base),
- (version, pkgarch))
- rows = data.fetchall()
- value = self._max_value(rows)
+ with closing(self.conn.cursor()) as cursor:
+ data = cursor.execute("SELECT value FROM %s where version=? AND pkgarch=? AND value LIKE '%s.%%';" % (self.table, base),
+ (version, pkgarch))
+ rows = data.fetchall()
+ value = self._max_value(rows)
- if value is not None:
- return increase_revision(value)
- else:
- return base + ".0"
+ if value is not None:
+ return increase_revision(value)
+ else:
+ return base + ".0"
def store_value(self, version, pkgarch, checksum, value):
"""Store new value in the database"""
- try:
- self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
- (version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
-
- self.dirty = True
+ with closing(self.conn.cursor()) as cursor:
+ try:
+ cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
+ (version, pkgarch, checksum, value))
+ except sqlite3.IntegrityError as exc:
+ logger.error(str(exc))
+ self.conn.commit()
def _get_value(self, version, pkgarch, checksum, history):
@@ -215,54 +190,56 @@ class PRTable(object):
return None
val = None
- data = self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
+ with closing(self.conn.cursor()) as cursor:
+ data = cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
(version, pkgarch, checksum))
- row = data.fetchone()
- if row is not None:
- val=row[0]
- else:
- #no value found, try to insert
- try:
- self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
- (version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
+ row = data.fetchone()
+ if row is not None:
+ val=row[0]
+ else:
+ #no value found, try to insert
+ try:
+ cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
+ (version, pkgarch, checksum, value))
+ except sqlite3.IntegrityError as exc:
+ logger.error(str(exc))
- self.dirty = True
+ self.conn.commit()
- data = self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
+ data = cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
(version, pkgarch, checksum))
- row = data.fetchone()
- if row is not None:
- val = row[0]
+ row = data.fetchone()
+ if row is not None:
+ val = row[0]
return val
def _import_no_hist(self, version, pkgarch, checksum, value):
if self.read_only:
return None
- try:
- #try to insert
- self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
- (version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- #already have the record, try to update
+ with closing(self.conn.cursor()) as cursor:
try:
- self._execute("UPDATE %s SET value=? WHERE version=? AND pkgarch=? AND checksum=? AND value<?"
- % (self.table),
- (value, version, pkgarch, checksum, value))
+ #try to insert
+ cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
+ (version, pkgarch, checksum, value))
except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
+ #already have the record, try to update
+ try:
+ cursor.execute("UPDATE %s SET value=? WHERE version=? AND pkgarch=? AND checksum=? AND value<?"
+ % (self.table),
+ (value, version, pkgarch, checksum, value))
+ except sqlite3.IntegrityError as exc:
+ logger.error(str(exc))
- self.dirty = True
+ self.conn.commit()
- data = self._execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=? AND value>=?;" % self.table,
+ data = cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=? AND value>=?;" % self.table,
(version, pkgarch, checksum, value))
- row=data.fetchone()
- if row is not None:
- return row[0]
- else:
- return None
+ row=data.fetchone()
+ if row is not None:
+ return row[0]
+ else:
+ return None
def importone(self, version, pkgarch, checksum, value, history=False):
if history:
@@ -272,56 +249,57 @@ class PRTable(object):
def export(self, version, pkgarch, checksum, colinfo, history=False):
metainfo = {}
- #column info
- if colinfo:
- metainfo["tbl_name"] = self.table
- metainfo["core_ver"] = prserv.__version__
- metainfo["col_info"] = []
- data = self._execute("PRAGMA table_info(%s);" % self.table)
+ with closing(self.conn.cursor()) as cursor:
+ #column info
+ if colinfo:
+ metainfo["tbl_name"] = self.table
+ metainfo["core_ver"] = prserv.__version__
+ metainfo["col_info"] = []
+ data = cursor.execute("PRAGMA table_info(%s);" % self.table)
+ for row in data:
+ col = {}
+ col["name"] = row["name"]
+ col["type"] = row["type"]
+ col["notnull"] = row["notnull"]
+ col["dflt_value"] = row["dflt_value"]
+ col["pk"] = row["pk"]
+ metainfo["col_info"].append(col)
+
+ #data info
+ datainfo = []
+
+ if history:
+ sqlstmt = "SELECT * FROM %s as T1 WHERE 1=1 " % self.table
+ else:
+ sqlstmt = "SELECT T1.version, T1.pkgarch, T1.checksum, T1.value FROM %s as T1, \
+ (SELECT version, pkgarch, max(value) as maxvalue FROM %s GROUP BY version, pkgarch) as T2 \
+ WHERE T1.version=T2.version AND T1.pkgarch=T2.pkgarch AND T1.value=T2.maxvalue " % (self.table, self.table)
+ sqlarg = []
+ where = ""
+ if version:
+ where += "AND T1.version=? "
+ sqlarg.append(str(version))
+ if pkgarch:
+ where += "AND T1.pkgarch=? "
+ sqlarg.append(str(pkgarch))
+ if checksum:
+ where += "AND T1.checksum=? "
+ sqlarg.append(str(checksum))
+
+ sqlstmt += where + ";"
+
+ if len(sqlarg):
+ data = cursor.execute(sqlstmt, tuple(sqlarg))
+ else:
+ data = cursor.execute(sqlstmt)
for row in data:
- col = {}
- col["name"] = row["name"]
- col["type"] = row["type"]
- col["notnull"] = row["notnull"]
- col["dflt_value"] = row["dflt_value"]
- col["pk"] = row["pk"]
- metainfo["col_info"].append(col)
-
- #data info
- datainfo = []
-
- if history:
- sqlstmt = "SELECT * FROM %s as T1 WHERE 1=1 " % self.table
- else:
- sqlstmt = "SELECT T1.version, T1.pkgarch, T1.checksum, T1.value FROM %s as T1, \
- (SELECT version, pkgarch, max(value) as maxvalue FROM %s GROUP BY version, pkgarch) as T2 \
- WHERE T1.version=T2.version AND T1.pkgarch=T2.pkgarch AND T1.value=T2.maxvalue " % (self.table, self.table)
- sqlarg = []
- where = ""
- if version:
- where += "AND T1.version=? "
- sqlarg.append(str(version))
- if pkgarch:
- where += "AND T1.pkgarch=? "
- sqlarg.append(str(pkgarch))
- if checksum:
- where += "AND T1.checksum=? "
- sqlarg.append(str(checksum))
-
- sqlstmt += where + ";"
-
- if len(sqlarg):
- data = self._execute(sqlstmt, tuple(sqlarg))
- else:
- data = self._execute(sqlstmt)
- for row in data:
- if row["version"]:
- col = {}
- col["version"] = row["version"]
- col["pkgarch"] = row["pkgarch"]
- col["checksum"] = row["checksum"]
- col["value"] = row["value"]
- datainfo.append(col)
+ if row["version"]:
+ col = {}
+ col["version"] = row["version"]
+ col["pkgarch"] = row["pkgarch"]
+ col["checksum"] = row["checksum"]
+ col["value"] = row["value"]
+ datainfo.append(col)
return (metainfo, datainfo)
def dump_db(self, fd):
@@ -345,14 +323,15 @@ class PRData(object):
raise e
uri = "file:%s%s" % (self.filename, "?mode=ro" if self.read_only else "")
logger.debug("Opening PRServ database '%s'" % (uri))
- self.connection=sqlite3.connect(uri, uri=True, isolation_level="EXCLUSIVE", check_same_thread = False)
+ self.connection=sqlite3.connect(uri, uri=True)
self.connection.row_factory=sqlite3.Row
- if not self.read_only:
- self.connection.execute("pragma synchronous = off;")
- self.connection.execute("PRAGMA journal_mode = MEMORY;")
+ self.connection.execute("PRAGMA synchronous = OFF;")
+ self.connection.execute("PRAGMA journal_mode = WAL;")
+ self.connection.commit()
self._tables={}
def disconnect(self):
+ self.connection.commit()
self.connection.close()
def __getitem__(self, tblname):
@@ -370,3 +349,4 @@ class PRData(object):
del self._tables[tblname]
logger.info("drop table %s" % (tblname))
self.connection.execute("DROP TABLE IF EXISTS %s;" % tblname)
+ self.connection.commit()
diff --git a/lib/prserv/serv.py b/lib/prserv/serv.py
index 05573d06cc..fd673b1851 100644
--- a/lib/prserv/serv.py
+++ b/lib/prserv/serv.py
@@ -44,8 +44,6 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
except:
self.server.table.sync()
raise
- else:
- self.server.table.sync_if_dirty()
async def handle_test_pr(self, request):
'''Finds the PR value corresponding to the request. If not found, returns None and doesn't insert a new value'''
@@ -233,15 +231,9 @@ class PRServer(bb.asyncrpc.AsyncServer):
return tasks
async def stop(self):
- self.table.sync_if_dirty()
self.db.disconnect()
await super().stop()
- def signal_handler(self):
- super().signal_handler()
- if self.table:
- self.table.sync()
-
class PRServSingleton(object):
def __init__(self, dbfile, logfile, host, port, upstream):
self.dbfile = dbfile
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 5/8] prserv: avoid possible race condition in database code
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
` (3 preceding siblings ...)
2024-04-30 17:15 ` [PATCH v6 4/8] prserv: enable database sharing michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 6/8] prserv: store_value() improvements michael.opdenacker
` (3 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
Remove a possible race condition by allowing a read-only
server to create the PR table anyway. This avoids a failure
if both a read-only and read-write server try to access
an empty database at the same time.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Suggested-by: Joshua Watt <jpewhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
lib/prserv/db.py | 25 +++++++++++--------------
1 file changed, 11 insertions(+), 14 deletions(-)
diff --git a/lib/prserv/db.py b/lib/prserv/db.py
index f430586d73..79c9001bf5 100644
--- a/lib/prserv/db.py
+++ b/lib/prserv/db.py
@@ -30,21 +30,18 @@ class PRTable(object):
self.read_only = read_only
self.table = table
+ # Creating the table even if the server is read-only.
+ # This avoids a race condition if a shared database
+ # is accessed by a read-only server first.
+
with closing(self.conn.cursor()) as cursor:
- if self.read_only:
- table_exists = cursor.execute(
- "SELECT count(*) FROM sqlite_master \
- WHERE type='table' AND name='%s'" % (self.table))
- if not table_exists:
- raise prserv.NotFoundError
- else:
- cursor.execute("CREATE TABLE IF NOT EXISTS %s \
- (version TEXT NOT NULL, \
- pkgarch TEXT NOT NULL, \
- checksum TEXT NOT NULL, \
- value TEXT, \
- PRIMARY KEY (version, pkgarch, checksum, value));" % self.table)
- self.conn.commit()
+ cursor.execute("CREATE TABLE IF NOT EXISTS %s \
+ (version TEXT NOT NULL, \
+ pkgarch TEXT NOT NULL, \
+ checksum TEXT NOT NULL, \
+ value TEXT, \
+ PRIMARY KEY (version, pkgarch, checksum, value));" % self.table)
+ self.conn.commit()
def _extremum_value(self, rows, is_max):
value = None
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 6/8] prserv: store_value() improvements
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
` (4 preceding siblings ...)
2024-04-30 17:15 ` [PATCH v6 5/8] prserv: avoid possible race condition in database code michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 7/8] prserv: import simplification michael.opdenacker
` (2 subsequent siblings)
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
Add a test_checksum_value() to test whether
a (version, pkgarch, checksum, value) entry already
exists in the database.
This is used to protect the store_value() function from
an error when trying to store a duplicate entry in the database.
Also check whether the current database is open in read-only mode.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
lib/prserv/db.py | 22 ++++++++++++++++------
1 file changed, 16 insertions(+), 6 deletions(-)
diff --git a/lib/prserv/db.py b/lib/prserv/db.py
index 79c9001bf5..88ed8e2125 100644
--- a/lib/prserv/db.py
+++ b/lib/prserv/db.py
@@ -78,6 +78,18 @@ class PRTable(object):
else:
return False
+ def test_checksum_value(self, version, pkgarch, checksum, value):
+ """Returns whether the specified value is found in the database for the specified package, architecture and checksum"""
+
+ with closing(self.conn.cursor()) as cursor:
+ data=cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? and checksum=? and value=?;" % self.table,
+ (version, pkgarch, checksum, value))
+ row=data.fetchone()
+ if row is not None:
+ return True
+ else:
+ return False
+
def test_value(self, version, pkgarch, value):
"""Returns whether the specified value is found in the database for the specified package and architecture"""
@@ -143,15 +155,13 @@ class PRTable(object):
return base + ".0"
def store_value(self, version, pkgarch, checksum, value):
- """Store new value in the database"""
+ """Store value in the database"""
- with closing(self.conn.cursor()) as cursor:
- try:
+ if not self.read_only and not self.test_checksum_value(version, pkgarch, checksum, value):
+ with closing(self.conn.cursor()) as cursor:
cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
(version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
- self.conn.commit()
+ self.conn.commit()
def _get_value(self, version, pkgarch, checksum, history):
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 7/8] prserv: import simplification
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
` (5 preceding siblings ...)
2024-04-30 17:15 ` [PATCH v6 6/8] prserv: store_value() improvements michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 8/8] prserv: add bitbake selftests michael.opdenacker
2024-04-30 20:31 ` [bitbake-devel] [PATCH v6 0/8] prserv: add support for an "upstream" server Jan-Simon Moeller
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
Simplify the importone() hook:
- to make it independent from the "history" mode which is
client specific.
- remove the "history" parameter
- we want all values to be imported for binary
reproducibility purposes.
- using the store_value() function (which warrants
you don't save the same value twice and doesn't write
when you're using a read-only server) is enough.
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
lib/prserv/client.py | 4 +--
lib/prserv/db.py | 64 +++-----------------------------------------
lib/prserv/serv.py | 3 +--
3 files changed, 6 insertions(+), 65 deletions(-)
diff --git a/lib/prserv/client.py b/lib/prserv/client.py
index 565c6f3872..9f5794c433 100644
--- a/lib/prserv/client.py
+++ b/lib/prserv/client.py
@@ -42,9 +42,9 @@ class PRAsyncClient(bb.asyncrpc.AsyncClient):
if response:
return response["value"]
- async def importone(self, version, pkgarch, checksum, value, history=False):
+ async def importone(self, version, pkgarch, checksum, value):
response = await self.invoke(
- {"import-one": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "value": value, "history": history}}
+ {"import-one": {"version": version, "pkgarch": pkgarch, "checksum": checksum, "value": value}}
)
if response:
return response["value"]
diff --git a/lib/prserv/db.py b/lib/prserv/db.py
index 88ed8e2125..2da493ddf5 100644
--- a/lib/prserv/db.py
+++ b/lib/prserv/db.py
@@ -192,67 +192,9 @@ class PRTable(object):
self.store_value(version, pkgarch, checksum, value)
return value
- def _import_hist(self, version, pkgarch, checksum, value):
- if self.read_only:
- return None
-
- val = None
- with closing(self.conn.cursor()) as cursor:
- data = cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
- (version, pkgarch, checksum))
- row = data.fetchone()
- if row is not None:
- val=row[0]
- else:
- #no value found, try to insert
- try:
- cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
- (version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
-
- self.conn.commit()
-
- data = cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=?;" % self.table,
- (version, pkgarch, checksum))
- row = data.fetchone()
- if row is not None:
- val = row[0]
- return val
-
- def _import_no_hist(self, version, pkgarch, checksum, value):
- if self.read_only:
- return None
-
- with closing(self.conn.cursor()) as cursor:
- try:
- #try to insert
- cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" % (self.table),
- (version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- #already have the record, try to update
- try:
- cursor.execute("UPDATE %s SET value=? WHERE version=? AND pkgarch=? AND checksum=? AND value<?"
- % (self.table),
- (value, version, pkgarch, checksum, value))
- except sqlite3.IntegrityError as exc:
- logger.error(str(exc))
-
- self.conn.commit()
-
- data = cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND checksum=? AND value>=?;" % self.table,
- (version, pkgarch, checksum, value))
- row=data.fetchone()
- if row is not None:
- return row[0]
- else:
- return None
-
- def importone(self, version, pkgarch, checksum, value, history=False):
- if history:
- return self._import_hist(version, pkgarch, checksum, value)
- else:
- return self._import_no_hist(version, pkgarch, checksum, value)
+ def importone(self, version, pkgarch, checksum, value):
+ self.store_value(version, pkgarch, checksum, value)
+ return value
def export(self, version, pkgarch, checksum, colinfo, history=False):
metainfo = {}
diff --git a/lib/prserv/serv.py b/lib/prserv/serv.py
index fd673b1851..d3ee43dcb5 100644
--- a/lib/prserv/serv.py
+++ b/lib/prserv/serv.py
@@ -180,9 +180,8 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
pkgarch = request["pkgarch"]
checksum = request["checksum"]
value = request["value"]
- history = request["history"]
- value = self.server.table.importone(version, pkgarch, checksum, value, history)
+ value = self.server.table.importone(version, pkgarch, checksum, value)
if value is not None:
response = {"value": value}
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH v6 8/8] prserv: add bitbake selftests
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
` (6 preceding siblings ...)
2024-04-30 17:15 ` [PATCH v6 7/8] prserv: import simplification michael.opdenacker
@ 2024-04-30 17:15 ` michael.opdenacker
2024-04-30 20:31 ` [bitbake-devel] [PATCH v6 0/8] prserv: add support for an "upstream" server Jan-Simon Moeller
8 siblings, 0 replies; 11+ messages in thread
From: michael.opdenacker @ 2024-04-30 17:15 UTC (permalink / raw)
To: bitbake-devel
Cc: Michael Opdenacker, Joshua Watt, Tim Orling, Thomas Petazzoni
From: Michael Opdenacker <michael.opdenacker@bootlin.com>
Run them with "bitbake-selftest prserv.tests"
Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
Cc: Joshua Watt <JPEWhacker@gmail.com>
Cc: Tim Orling <ticotimo@gmail.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
---
bin/bitbake-selftest | 2 +
lib/prserv/tests.py | 386 +++++++++++++++++++++++++++++++++++++++++++
2 files changed, 388 insertions(+)
create mode 100644 bitbake/lib/prserv/tests.py
diff --git a/bin/bitbake-selftest b/bin/bitbake-selftest
index f25f23b1ae..ce901232fe 100755
--- a/bin/bitbake-selftest
+++ b/bin/bitbake-selftest
@@ -15,6 +15,7 @@ import unittest
try:
import bb
import hashserv
+ import prserv
import layerindexlib
except RuntimeError as exc:
sys.exit(str(exc))
@@ -33,6 +34,7 @@ tests = ["bb.tests.codeparser",
"bb.tests.utils",
"bb.tests.compression",
"hashserv.tests",
+ "prserv.tests",
"layerindexlib.tests.layerindexobj",
"layerindexlib.tests.restapi",
"layerindexlib.tests.cooker"]
diff --git a/lib/prserv/tests.py b/lib/prserv/tests.py
new file mode 100644
index 0000000000..8765b129f2
--- /dev/null
+++ b/lib/prserv/tests.py
@@ -0,0 +1,386 @@
+#! /usr/bin/env python3
+#
+# Copyright (C) 2024 BitBake Contributors
+#
+# SPDX-License-Identifier: GPL-2.0-only
+#
+
+from . import create_server, create_client, increase_revision, revision_greater, revision_smaller, _revision_greater_or_equal
+import prserv.db as db
+from bb.asyncrpc import InvokeError
+import logging
+import os
+import sys
+import tempfile
+import unittest
+import socket
+import subprocess
+from pathlib import Path
+
+THIS_DIR = Path(__file__).parent
+BIN_DIR = THIS_DIR.parent.parent / "bin"
+
+version = "dummy-1.0-r0"
+pkgarch = "core2-64"
+other_arch = "aarch64"
+
+checksumX = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4f0"
+checksum0 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a0"
+checksum1 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a1"
+checksum2 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a2"
+checksum3 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a3"
+checksum4 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a4"
+checksum5 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a5"
+checksum6 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a6"
+checksum7 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a7"
+checksum8 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a8"
+checksum9 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4a9"
+checksum10 = "51bf8189dbe9ea81fa6dd89608bf19380c437a9cf12f6c6239887801ba4ab4aa"
+
+def server_prefunc(server, name):
+ logging.basicConfig(level=logging.DEBUG, filename='prserv-%s.log' % name, filemode='w',
+ format='%(levelname)s %(filename)s:%(lineno)d %(message)s')
+ server.logger.debug("Running server %s" % name)
+ sys.stdout = open('prserv-stdout-%s.log' % name, 'w')
+ sys.stderr = sys.stdout
+
+class PRTestSetup(object):
+
+ def start_server(self, name, dbfile, upstream=None, read_only=False, prefunc=server_prefunc):
+
+ def cleanup_server(server):
+ if server.process.exitcode is not None:
+ return
+ server.process.terminate()
+ server.process.join()
+
+ server = create_server(socket.gethostbyname("localhost") + ":0",
+ dbfile,
+ upstream=upstream,
+ read_only=read_only)
+
+ server.serve_as_process(prefunc=prefunc, args=(name,))
+ self.addCleanup(cleanup_server, server)
+
+ return server
+
+ def start_client(self, server_address):
+ def cleanup_client(client):
+ client.close()
+
+ client = create_client(server_address)
+ self.addCleanup(cleanup_client, client)
+
+ return client
+
+class FunctionTests(unittest.TestCase):
+
+ def setUp(self):
+ self.temp_dir = tempfile.TemporaryDirectory(prefix='bb-prserv')
+ self.addCleanup(self.temp_dir.cleanup)
+
+ def test_increase_revision(self):
+ self.assertEqual(increase_revision("1"), "2")
+ self.assertEqual(increase_revision("1.0"), "1.1")
+ self.assertEqual(increase_revision("1.1.1"), "1.1.2")
+ self.assertEqual(increase_revision("1.1.1.3"), "1.1.1.4")
+ self.assertRaises(ValueError, increase_revision, "1.a")
+ self.assertRaises(ValueError, increase_revision, "1.")
+ self.assertRaises(ValueError, increase_revision, "")
+
+ def test_revision_greater_or_equal(self):
+ self.assertTrue(_revision_greater_or_equal("2", "2"))
+ self.assertTrue(_revision_greater_or_equal("2", "1"))
+ self.assertTrue(_revision_greater_or_equal("10", "2"))
+ self.assertTrue(_revision_greater_or_equal("1.10", "1.2"))
+ self.assertFalse(_revision_greater_or_equal("1.2", "1.10"))
+ self.assertTrue(_revision_greater_or_equal("1.10", "1"))
+ self.assertTrue(_revision_greater_or_equal("1.10.1", "1.10"))
+ self.assertFalse(_revision_greater_or_equal("1.10.1", "1.10.2"))
+ self.assertTrue(_revision_greater_or_equal("1.10.1", "1.10.1"))
+ self.assertTrue(_revision_greater_or_equal("1.10.1", "1"))
+ self.assertTrue(revision_greater("1.20", "1.3"))
+ self.assertTrue(revision_smaller("1.3", "1.20"))
+
+ # DB tests
+
+ def test_db(self):
+ dbfile = os.path.join(self.temp_dir.name, "testtable.sqlite3")
+
+ self.db = db.PRData(dbfile)
+ self.table = self.db["PRMAIN"]
+
+ self.table.store_value(version, pkgarch, checksum0, "0")
+ self.table.store_value(version, pkgarch, checksum1, "1")
+ # "No history" mode supports multiple PRs for the same checksum
+ self.table.store_value(version, pkgarch, checksum0, "2")
+ self.table.store_value(version, pkgarch, checksum2, "1.0")
+
+ self.assertTrue(self.table.test_package(version, pkgarch))
+ self.assertFalse(self.table.test_package(version, other_arch))
+
+ self.assertTrue(self.table.test_value(version, pkgarch, "0"))
+ self.assertTrue(self.table.test_value(version, pkgarch, "1"))
+ self.assertTrue(self.table.test_value(version, pkgarch, "2"))
+
+ self.assertEqual(self.table.find_package_max_value(version, pkgarch), "2")
+
+ self.assertEqual(self.table.find_min_value(version, pkgarch, checksum0), "0")
+ self.assertEqual(self.table.find_max_value(version, pkgarch, checksum0), "2")
+
+ # Test history modes
+ self.assertEqual(self.table.find_value(version, pkgarch, checksum0, True), "0")
+ self.assertEqual(self.table.find_value(version, pkgarch, checksum0, False), "2")
+
+ self.assertEqual(self.table.find_new_subvalue(version, pkgarch, "3"), "3.0")
+ self.assertEqual(self.table.find_new_subvalue(version, pkgarch, "1"), "1.1")
+
+ # Revision comparison tests
+ self.table.store_value(version, pkgarch, checksum1, "1.3")
+ self.table.store_value(version, pkgarch, checksum1, "1.20")
+ self.assertEqual(self.table.find_min_value(version, pkgarch, checksum1), "1")
+ self.assertEqual(self.table.find_max_value(version, pkgarch, checksum1), "1.20")
+
+class PRBasicTests(PRTestSetup, unittest.TestCase):
+
+ def setUp(self):
+ self.temp_dir = tempfile.TemporaryDirectory(prefix='bb-prserv')
+ self.addCleanup(self.temp_dir.cleanup)
+
+ dbfile = os.path.join(self.temp_dir.name, "prtest-basic.sqlite3")
+
+ self.server1 = self.start_server("basic", dbfile)
+ self.client1 = self.start_client(self.server1.address)
+
+ def test_basic(self):
+
+ # Checks on non existing configuration
+
+ result = self.client1.test_pr(version, pkgarch, checksum0)
+ self.assertIsNone(result, "test_pr should return 'None' for a non existing PR")
+
+ result = self.client1.test_package(version, pkgarch)
+ self.assertFalse(result, "test_package should return 'False' for a non existing PR")
+
+ result = self.client1.max_package_pr(version, pkgarch)
+ self.assertIsNone(result, "max_package_pr should return 'None' for a non existing PR")
+
+ # Add a first configuration
+
+ result = self.client1.getPR(version, pkgarch, checksum0)
+ self.assertEqual(result, "0", "getPR: initial PR of a package should be '0'")
+
+ result = self.client1.test_pr(version, pkgarch, checksum0)
+ self.assertEqual(result, "0", "test_pr should return '0' here, matching the result of getPR")
+
+ result = self.client1.test_package(version, pkgarch)
+ self.assertTrue(result, "test_package should return 'True' for an existing PR")
+
+ result = self.client1.max_package_pr(version, pkgarch)
+ self.assertEqual(result, "0", "max_package_pr should return '0' in the current test series")
+
+ # Check that the same request gets the same value
+
+ result = self.client1.getPR(version, pkgarch, checksum0)
+ self.assertEqual(result, "0", "getPR: asking for the same PR a second time in a row should return the same value.")
+
+ # Add new configurations
+
+ result = self.client1.getPR(version, pkgarch, checksum1)
+ self.assertEqual(result, "1", "getPR: second PR of a package should be '1'")
+
+ result = self.client1.test_pr(version, pkgarch, checksum1)
+ self.assertEqual(result, "1", "test_pr should return '1' here, matching the result of getPR")
+
+ result = self.client1.max_package_pr(version, pkgarch)
+ self.assertEqual(result, "1", "max_package_pr should return '1' in the current test series")
+
+ result = self.client1.getPR(version, pkgarch, checksum2)
+ self.assertEqual(result, "2", "getPR: second PR of a package should be '2'")
+
+ result = self.client1.test_pr(version, pkgarch, checksum2)
+ self.assertEqual(result, "2", "test_pr should return '2' here, matching the result of getPR")
+
+ result = self.client1.max_package_pr(version, pkgarch)
+ self.assertEqual(result, "2", "max_package_pr should return '2' in the current test series")
+
+ result = self.client1.getPR(version, pkgarch, checksum3)
+ self.assertEqual(result, "3", "getPR: second PR of a package should be '3'")
+
+ result = self.client1.test_pr(version, pkgarch, checksum3)
+ self.assertEqual(result, "3", "test_pr should return '3' here, matching the result of getPR")
+
+ result = self.client1.max_package_pr(version, pkgarch)
+ self.assertEqual(result, "3", "max_package_pr should return '3' in the current test series")
+
+ # Ask again for the first configuration
+
+ result = self.client1.getPR(version, pkgarch, checksum0)
+ self.assertEqual(result, "4", "getPR: should return '4' in this configuration")
+
+ # Ask again with explicit "no history" mode
+
+ result = self.client1.getPR(version, pkgarch, checksum0, False)
+ self.assertEqual(result, "4", "getPR: should return '4' in this configuration")
+
+ # Ask again with explicit "history" mode. This should return the first recorded PR for checksum0
+
+ result = self.client1.getPR(version, pkgarch, checksum0, True)
+ self.assertEqual(result, "0", "getPR: should return '0' in this configuration")
+
+ # Check again that another pkgarg resets the counters
+
+ result = self.client1.test_pr(version, other_arch, checksum0)
+ self.assertIsNone(result, "test_pr should return 'None' for a non existing PR")
+
+ result = self.client1.test_package(version, other_arch)
+ self.assertFalse(result, "test_package should return 'False' for a non existing PR")
+
+ result = self.client1.max_package_pr(version, other_arch)
+ self.assertIsNone(result, "max_package_pr should return 'None' for a non existing PR")
+
+ # Now add the configuration
+
+ result = self.client1.getPR(version, other_arch, checksum0)
+ self.assertEqual(result, "0", "getPR: initial PR of a package should be '0'")
+
+ result = self.client1.test_pr(version, other_arch, checksum0)
+ self.assertEqual(result, "0", "test_pr should return '0' here, matching the result of getPR")
+
+ result = self.client1.test_package(version, other_arch)
+ self.assertTrue(result, "test_package should return 'True' for an existing PR")
+
+ result = self.client1.max_package_pr(version, other_arch)
+ self.assertEqual(result, "0", "max_package_pr should return '0' in the current test series")
+
+ result = self.client1.is_readonly()
+ self.assertFalse(result, "Server should not be described as 'read-only'")
+
+class PRUpstreamTests(PRTestSetup, unittest.TestCase):
+
+ def setUp(self):
+
+ self.temp_dir = tempfile.TemporaryDirectory(prefix='bb-prserv')
+ self.addCleanup(self.temp_dir.cleanup)
+
+ dbfile2 = os.path.join(self.temp_dir.name, "prtest-upstream2.sqlite3")
+ self.server2 = self.start_server("upstream2", dbfile2)
+ self.client2 = self.start_client(self.server2.address)
+
+ dbfile1 = os.path.join(self.temp_dir.name, "prtest-upstream1.sqlite3")
+ self.server1 = self.start_server("upstream1", dbfile1, upstream=self.server2.address)
+ self.client1 = self.start_client(self.server1.address)
+
+ dbfile0 = os.path.join(self.temp_dir.name, "prtest-local.sqlite3")
+ self.server0 = self.start_server("local", dbfile0, upstream=self.server1.address)
+ self.client0 = self.start_client(self.server0.address)
+ self.shared_db = dbfile0
+
+ def test_upstream_and_readonly(self):
+
+ # For identical checksums, all servers should return the same PR
+
+ result = self.client2.getPR(version, pkgarch, checksum0)
+ self.assertEqual(result, "0", "getPR: initial PR of a package should be '0'")
+
+ result = self.client1.getPR(version, pkgarch, checksum0)
+ self.assertEqual(result, "0", "getPR: initial PR of a package should be '0' (same as upstream)")
+
+ result = self.client0.getPR(version, pkgarch, checksum0)
+ self.assertEqual(result, "0", "getPR: initial PR of a package should be '0' (same as upstream)")
+
+ # Now introduce new checksums on server1 for, same version
+
+ result = self.client1.getPR(version, pkgarch, checksum1)
+ self.assertEqual(result, "0.0", "getPR: first PR of a package which has a different checksum upstream should be '0.0'")
+
+ result = self.client1.getPR(version, pkgarch, checksum2)
+ self.assertEqual(result, "0.1", "getPR: second PR of a package that has a different checksum upstream should be '0.1'")
+
+ # Now introduce checksums on server0 for, same version
+
+ result = self.client1.getPR(version, pkgarch, checksum1)
+ self.assertEqual(result, "0.2", "getPR: can't decrease for known PR")
+
+ result = self.client1.getPR(version, pkgarch, checksum2)
+ self.assertEqual(result, "0.3")
+
+ result = self.client1.max_package_pr(version, pkgarch)
+ self.assertEqual(result, "0.3")
+
+ result = self.client0.getPR(version, pkgarch, checksum3)
+ self.assertEqual(result, "0.3.0", "getPR: first PR of a package that doesn't exist upstream should be '0.3.0'")
+
+ result = self.client0.getPR(version, pkgarch, checksum4)
+ self.assertEqual(result, "0.3.1", "getPR: second PR of a package that doesn't exist upstream should be '0.3.1'")
+
+ result = self.client0.getPR(version, pkgarch, checksum3)
+ self.assertEqual(result, "0.3.2")
+
+ # More upstream updates
+ # Here, we assume no communication between server2 and server0. server2 only impacts server0
+ # after impacting server1
+
+ self.assertEqual(self.client2.getPR(version, pkgarch, checksum5), "1")
+ self.assertEqual(self.client1.getPR(version, pkgarch, checksum6), "1.0")
+ self.assertEqual(self.client1.getPR(version, pkgarch, checksum7), "1.1")
+ self.assertEqual(self.client0.getPR(version, pkgarch, checksum8), "1.1.0")
+ self.assertEqual(self.client0.getPR(version, pkgarch, checksum9), "1.1.1")
+
+ # "history" mode tests
+
+ self.assertEqual(self.client2.getPR(version, pkgarch, checksum0, True), "0")
+ self.assertEqual(self.client1.getPR(version, pkgarch, checksum2, True), "0.1")
+ self.assertEqual(self.client0.getPR(version, pkgarch, checksum3, True), "0.3.0")
+
+ # More "no history" mode tests
+
+ self.assertEqual(self.client2.getPR(version, pkgarch, checksum0), "2")
+ self.assertEqual(self.client1.getPR(version, pkgarch, checksum0), "2") # Same as upstream
+ self.assertEqual(self.client0.getPR(version, pkgarch, checksum0), "2") # Same as upstream
+ self.assertEqual(self.client1.getPR(version, pkgarch, checksum7), "3") # This could be surprising, but since the previous revision was "2", increasing it yields "3".
+ # We don't know how many upstream servers we have
+ # Start read-only server with server1 as upstream
+ self.server_ro = self.start_server("local-ro", self.shared_db, upstream=self.server1.address, read_only=True)
+ self.client_ro = self.start_client(self.server_ro.address)
+
+ self.assertTrue(self.client_ro.is_readonly(), "Database should be described as 'read-only'")
+
+ # Checks on non existing configurations
+ self.assertIsNone(self.client_ro.test_pr(version, pkgarch, checksumX))
+ self.assertFalse(self.client_ro.test_package("unknown", pkgarch))
+
+ # Look up existing configurations
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum0), "3") # "no history" mode
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum0, True), "0") # "history" mode
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum3), "3")
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum3, True), "0.3.0")
+ self.assertEqual(self.client_ro.max_package_pr(version, pkgarch), "2") # normal as "3" was never saved
+
+ # Try to insert a new value. Here this one is know upstream.
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum7), "3")
+ # Try to insert a completely new value. As the max upstream value is already "3", it should be "3.0"
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum10), "3.0")
+ # Same with another value which only exists in the upstream upstream server
+ # This time, as the upstream server doesn't know it, it will ask its upstream server. So that's a known one.
+ self.assertEqual(self.client_ro.getPR(version, pkgarch, checksum9), "3")
+
+class ScriptTests(unittest.TestCase):
+
+ def setUp(self):
+
+ self.temp_dir = tempfile.TemporaryDirectory(prefix='bb-prserv')
+ self.addCleanup(self.temp_dir.cleanup)
+ self.dbfile = os.path.join(self.temp_dir.name, "prtest.sqlite3")
+
+ def test_1_start_bitbake_prserv(self):
+ try:
+ subprocess.check_call([BIN_DIR / "bitbake-prserv", "--start", "-f", self.dbfile])
+ except subprocess.CalledProcessError as e:
+ self.fail("Failed to start bitbake-prserv: %s" % e.returncode)
+
+ def test_2_stop_bitbake_prserv(self):
+ try:
+ subprocess.check_call([BIN_DIR / "bitbake-prserv", "--stop"])
+ except subprocess.CalledProcessError as e:
+ self.fail("Failed to stop bitbake-prserv: %s" % e.returncode)
--
2.34.1
^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [bitbake-devel] [PATCH v6 4/8] prserv: enable database sharing
2024-04-30 17:15 ` [PATCH v6 4/8] prserv: enable database sharing michael.opdenacker
@ 2024-04-30 20:29 ` Jan-Simon Moeller
0 siblings, 0 replies; 11+ messages in thread
From: Jan-Simon Moeller @ 2024-04-30 20:29 UTC (permalink / raw)
To: michael.opdenacker; +Cc: bitbake-devel
Reviewed-by: Jan-Siumon Möller <dl9pf@gmx.de>
Am Dienstag, 30. April 2024, 19:15:08 CEST schrieb Michael Opdenacker via
lists.openembedded.org:
> From: Michael Opdenacker <michael.opdenacker@bootlin.com>
>
> sqlite3 can allow multiple processes to access the database
> simultaneously, but it must be opened correctly. The key change is that
> the database is no longer opened in "exclusive" mode (defaulting to
> shared mode). In addition, the journal is set to "WAL" mode, as this is
> the most efficient for dealing with simultaneous access between
> different processes. In order to keep the database performance,
> synchronous mode is set to "off". The WAL journal will protect against
> incomplete transactions in any given client, however the database will
> not be protected against unexpected power loss from the OS (which is a
> fine trade off for performance, and also the same as the previous
> implementation).
>
> The use of a database cursor enabled to remove the _execute() wrapper.
> The cursor automatically makes sure that the query happens in an atomic
> transaction and commits when finished.
>
> This also removes the need for a "dirty" flag for the database and
> for explicit database syncing, which simplifies the code.
>
> Signed-off-by: Michael Opdenacker <michael.opdenacker@bootlin.com>
> Signed-off-by: Joshua Watt <JPEWhacker@gmail.com>
> Cc: Tim Orling <ticotimo@gmail.com>
> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> ---
> lib/prserv/db.py | 322 +++++++++++++++++++++------------------------
> lib/prserv/serv.py | 8 --
> 2 files changed, 151 insertions(+), 179 deletions(-)
>
> diff --git a/lib/prserv/db.py b/lib/prserv/db.py
> index b2520f3158..f430586d73 100644
> --- a/lib/prserv/db.py
> +++ b/lib/prserv/db.py
> @@ -8,21 +8,13 @@ import logging
> import os.path
> import errno
> import prserv
> -import time
> +import sqlite3
>
> +from contextlib import closing
> from . import increase_revision, revision_greater, revision_smaller
>
> -try:
> - import sqlite3
> -except ImportError:
> - from pysqlite2 import dbapi2 as sqlite3
> -
> logger = logging.getLogger("BitBake.PRserv")
>
> -sqlversion = sqlite3.sqlite_version_info
> -if sqlversion[0] < 3 or (sqlversion[0] == 3 and sqlversion[1] < 3):
> - raise Exception("sqlite3 version 3.3.0 or later is required.")
> -
> #
> # "No History" mode - for a given query tuple (version, pkgarch, checksum),
> # the returned value will be the largest among all the values of the same
> @@ -31,40 +23,28 @@ if sqlversion[0] < 3 or (sqlversion[0] == 3 and
> sqlversion[1] < 3): # "History" mode - Return a new higher value for
> previously unseen query # tuple (version, pkgarch, checksum), otherwise
> return historical value. # Value can decrement if returning to a previous
> build.
> -#
>
> class PRTable(object):
> def __init__(self, conn, table, read_only):
> self.conn = conn
> self.read_only = read_only
> - self.dirty = False
> self.table = table
>
> - if self.read_only:
> - table_exists = self._execute(
> - "SELECT count(*) FROM sqlite_master \
> - WHERE type='table' AND name='%s'" % (self.table))
> - if not table_exists:
> - raise prserv.NotFoundError
> - else:
> - self._execute("CREATE TABLE IF NOT EXISTS %s \
> - (version TEXT NOT NULL, \
> - pkgarch TEXT NOT NULL, \
> - checksum TEXT NOT NULL, \
> - value TEXT, \
> - PRIMARY KEY (version, pkgarch, checksum, value));"
> % self.table) -
> - def _execute(self, *query):
> - """Execute a query, waiting to acquire a lock if necessary"""
> - start = time.time()
> - end = start + 20
> - while True:
> - try:
> - return self.conn.execute(*query)
> - except sqlite3.OperationalError as exc:
> - if "is locked" in str(exc) and end > time.time():
> - continue
> - raise exc
> + with closing(self.conn.cursor()) as cursor:
> + if self.read_only:
> + table_exists = cursor.execute(
> + "SELECT count(*) FROM sqlite_master \
> + WHERE type='table' AND name='%s'" %
> (self.table)) + if not table_exists:
> + raise prserv.NotFoundError
> + else:
> + cursor.execute("CREATE TABLE IF NOT EXISTS %s \
> + (version TEXT NOT NULL, \
> + pkgarch TEXT NOT NULL, \
> + checksum TEXT NOT NULL, \
> + value TEXT, \
> + PRIMARY KEY (version, pkgarch, checksum,
> value));" % self.table) + self.conn.commit()
>
> def _extremum_value(self, rows, is_max):
> value = None
> @@ -88,49 +68,42 @@ class PRTable(object):
> def _min_value(self, rows):
> return self._extremum_value(rows, False)
>
> - def sync(self):
> - if not self.read_only:
> - self.conn.commit()
> - self._execute("BEGIN EXCLUSIVE TRANSACTION")
> -
> - def sync_if_dirty(self):
> - if self.dirty:
> - self.sync()
> - self.dirty = False
> -
> def test_package(self, version, pkgarch):
> """Returns whether the specified package version is found in the
> database for the specified architecture"""
>
> # Just returns the value if found or None otherwise
> - data=self._execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=?;" % self.table, - (version, pkgarch))
> - row=data.fetchone()
> - if row is not None:
> - return True
> - else:
> - return False
> + with closing(self.conn.cursor()) as cursor:
> + data=cursor.execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=?;" % self.table, + (version,
> pkgarch))
> + row=data.fetchone()
> + if row is not None:
> + return True
> + else:
> + return False
>
> def test_value(self, version, pkgarch, value):
> """Returns whether the specified value is found in the database for
> the specified package and architecture"""
>
> # Just returns the value if found or None otherwise
> - data=self._execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=? and value=?;" % self.table, - (version,
> pkgarch, value))
> - row=data.fetchone()
> - if row is not None:
> - return True
> - else:
> - return False
> + with closing(self.conn.cursor()) as cursor:
> + data=cursor.execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=? and value=?;" % self.table, +
> (version, pkgarch, value))
> + row=data.fetchone()
> + if row is not None:
> + return True
> + else:
> + return False
>
>
> def find_package_max_value(self, version, pkgarch):
> """Returns the greatest value for (version, pkgarch), or None if
> not found. Doesn't create a new value"""
>
> - data = self._execute("SELECT value FROM %s where version=? AND
> pkgarch=?;" % (self.table), - (version,
> pkgarch))
> - rows = data.fetchall()
> - value = self._max_value(rows)
> - return value
> + with closing(self.conn.cursor()) as cursor:
> + data = cursor.execute("SELECT value FROM %s where version=? AND
> pkgarch=?;" % (self.table), + (version,
> pkgarch))
> + rows = data.fetchall()
> + value = self._max_value(rows)
> + return value
>
> def find_value(self, version, pkgarch, checksum, history=False):
> """Returns the value for the specified checksum if found or None
> otherwise.""" @@ -145,10 +118,11 @@ class PRTable(object):
> """Returns the maximum (if is_max is True) or minimum (if is_max is
> False) value for (version, pkgarch, checksum), or None if not found.
> Doesn't create a new value"""
>
> - data = self._execute("SELECT value FROM %s where version=? AND
> pkgarch=? AND checksum=?;" % (self.table), -
> (version, pkgarch, checksum))
> - rows = data.fetchall()
> - return self._extremum_value(rows, is_max)
> + with closing(self.conn.cursor()) as cursor:
> + data = cursor.execute("SELECT value FROM %s where version=? AND
> pkgarch=? AND checksum=?;" % (self.table), +
> (version, pkgarch, checksum))
> + rows = data.fetchall()
> + return self._extremum_value(rows, is_max)
>
> def find_max_value(self, version, pkgarch, checksum):
> return self._find_extremum_value(version, pkgarch, checksum, True)
> @@ -160,26 +134,27 @@ class PRTable(object):
> """Take and increase the greatest "<base>.y" value for (version,
> pkgarch), or return "<base>.0" if not found. This doesn't store a new
> value."""
>
> - data = self._execute("SELECT value FROM %s where version=? AND
> pkgarch=? AND value LIKE '%s.%%';" % (self.table, base), -
> (version, pkgarch))
> - rows = data.fetchall()
> - value = self._max_value(rows)
> + with closing(self.conn.cursor()) as cursor:
> + data = cursor.execute("SELECT value FROM %s where version=? AND
> pkgarch=? AND value LIKE '%s.%%';" % (self.table, base), +
> (version, pkgarch))
> + rows = data.fetchall()
> + value = self._max_value(rows)
>
> - if value is not None:
> - return increase_revision(value)
> - else:
> - return base + ".0"
> + if value is not None:
> + return increase_revision(value)
> + else:
> + return base + ".0"
>
> def store_value(self, version, pkgarch, checksum, value):
> """Store new value in the database"""
>
> - try:
> - self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" %
> (self.table), - (version, pkgarch, checksum, value))
> - except sqlite3.IntegrityError as exc:
> - logger.error(str(exc))
> -
> - self.dirty = True
> + with closing(self.conn.cursor()) as cursor:
> + try:
> + cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" %
> (self.table), + (version, pkgarch, checksum,
> value))
> + except sqlite3.IntegrityError as exc:
> + logger.error(str(exc))
> + self.conn.commit()
>
> def _get_value(self, version, pkgarch, checksum, history):
>
> @@ -215,54 +190,56 @@ class PRTable(object):
> return None
>
> val = None
> - data = self._execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=? AND checksum=?;" % self.table, + with
> closing(self.conn.cursor()) as cursor:
> + data = cursor.execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=? AND checksum=?;" % self.table, (version, pkgarch, checksum))
> - row = data.fetchone()
> - if row is not None:
> - val=row[0]
> - else:
> - #no value found, try to insert
> - try:
> - self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" %
> (self.table), - (version, pkgarch, checksum,
> value))
> - except sqlite3.IntegrityError as exc:
> - logger.error(str(exc))
> + row = data.fetchone()
> + if row is not None:
> + val=row[0]
> + else:
> + #no value found, try to insert
> + try:
> + cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" %
> (self.table), + (version, pkgarch, checksum,
> value)) + except sqlite3.IntegrityError as exc:
> + logger.error(str(exc))
>
> - self.dirty = True
> + self.conn.commit()
>
> - data = self._execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=? AND checksum=?;" % self.table, + data =
> cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND
> checksum=?;" % self.table, (version, pkgarch, checksum))
> - row = data.fetchone()
> - if row is not None:
> - val = row[0]
> + row = data.fetchone()
> + if row is not None:
> + val = row[0]
> return val
>
> def _import_no_hist(self, version, pkgarch, checksum, value):
> if self.read_only:
> return None
>
> - try:
> - #try to insert
> - self._execute("INSERT INTO %s VALUES (?, ?, ?, ?);" %
> (self.table), - (version, pkgarch, checksum,
> value))
> - except sqlite3.IntegrityError as exc:
> - #already have the record, try to update
> + with closing(self.conn.cursor()) as cursor:
> try:
> - self._execute("UPDATE %s SET value=? WHERE version=? AND
> pkgarch=? AND checksum=? AND value<?" - %
> (self.table),
> - (value, version, pkgarch, checksum, value))
> + #try to insert
> + cursor.execute("INSERT INTO %s VALUES (?, ?, ?, ?);" %
> (self.table), + (version, pkgarch, checksum,
> value)) except sqlite3.IntegrityError as exc:
> - logger.error(str(exc))
> + #already have the record, try to update
> + try:
> + cursor.execute("UPDATE %s SET value=? WHERE version=?
> AND pkgarch=? AND checksum=? AND value<?" +
> % (self.table),
> + (value, version, pkgarch, checksum,
> value)) + except sqlite3.IntegrityError as exc:
> + logger.error(str(exc))
>
> - self.dirty = True
> + self.conn.commit()
>
> - data = self._execute("SELECT value FROM %s WHERE version=? AND
> pkgarch=? AND checksum=? AND value>=?;" % self.table, + data =
> cursor.execute("SELECT value FROM %s WHERE version=? AND pkgarch=? AND
> checksum=? AND value>=?;" % self.table, (version, pkgarch, checksum,
> value))
> - row=data.fetchone()
> - if row is not None:
> - return row[0]
> - else:
> - return None
> + row=data.fetchone()
> + if row is not None:
> + return row[0]
> + else:
> + return None
>
> def importone(self, version, pkgarch, checksum, value, history=False):
> if history:
> @@ -272,56 +249,57 @@ class PRTable(object):
>
> def export(self, version, pkgarch, checksum, colinfo, history=False):
> metainfo = {}
> - #column info
> - if colinfo:
> - metainfo["tbl_name"] = self.table
> - metainfo["core_ver"] = prserv.__version__
> - metainfo["col_info"] = []
> - data = self._execute("PRAGMA table_info(%s);" % self.table)
> + with closing(self.conn.cursor()) as cursor:
> + #column info
> + if colinfo:
> + metainfo["tbl_name"] = self.table
> + metainfo["core_ver"] = prserv.__version__
> + metainfo["col_info"] = []
> + data = cursor.execute("PRAGMA table_info(%s);" %
> self.table) + for row in data:
> + col = {}
> + col["name"] = row["name"]
> + col["type"] = row["type"]
> + col["notnull"] = row["notnull"]
> + col["dflt_value"] = row["dflt_value"]
> + col["pk"] = row["pk"]
> + metainfo["col_info"].append(col)
> +
> + #data info
> + datainfo = []
> +
> + if history:
> + sqlstmt = "SELECT * FROM %s as T1 WHERE 1=1 " % self.table
> + else:
> + sqlstmt = "SELECT T1.version, T1.pkgarch, T1.checksum,
> T1.value FROM %s as T1, \ + (SELECT version,
> pkgarch, max(value) as maxvalue FROM %s GROUP BY version, pkgarch) as T2 \
> + WHERE T1.version=T2.version AND
> T1.pkgarch=T2.pkgarch AND T1.value=T2.maxvalue " % (self.table, self.table)
> + sqlarg = []
> + where = ""
> + if version:
> + where += "AND T1.version=? "
> + sqlarg.append(str(version))
> + if pkgarch:
> + where += "AND T1.pkgarch=? "
> + sqlarg.append(str(pkgarch))
> + if checksum:
> + where += "AND T1.checksum=? "
> + sqlarg.append(str(checksum))
> +
> + sqlstmt += where + ";"
> +
> + if len(sqlarg):
> + data = cursor.execute(sqlstmt, tuple(sqlarg))
> + else:
> + data = cursor.execute(sqlstmt)
> for row in data:
> - col = {}
> - col["name"] = row["name"]
> - col["type"] = row["type"]
> - col["notnull"] = row["notnull"]
> - col["dflt_value"] = row["dflt_value"]
> - col["pk"] = row["pk"]
> - metainfo["col_info"].append(col)
> -
> - #data info
> - datainfo = []
> -
> - if history:
> - sqlstmt = "SELECT * FROM %s as T1 WHERE 1=1 " % self.table
> - else:
> - sqlstmt = "SELECT T1.version, T1.pkgarch, T1.checksum, T1.value
> FROM %s as T1, \ - (SELECT version, pkgarch, max(value)
> as maxvalue FROM %s GROUP BY version, pkgarch) as T2 \ -
> WHERE T1.version=T2.version AND T1.pkgarch=T2.pkgarch AND
> T1.value=T2.maxvalue " % (self.table, self.table) - sqlarg = []
> - where = ""
> - if version:
> - where += "AND T1.version=? "
> - sqlarg.append(str(version))
> - if pkgarch:
> - where += "AND T1.pkgarch=? "
> - sqlarg.append(str(pkgarch))
> - if checksum:
> - where += "AND T1.checksum=? "
> - sqlarg.append(str(checksum))
> -
> - sqlstmt += where + ";"
> -
> - if len(sqlarg):
> - data = self._execute(sqlstmt, tuple(sqlarg))
> - else:
> - data = self._execute(sqlstmt)
> - for row in data:
> - if row["version"]:
> - col = {}
> - col["version"] = row["version"]
> - col["pkgarch"] = row["pkgarch"]
> - col["checksum"] = row["checksum"]
> - col["value"] = row["value"]
> - datainfo.append(col)
> + if row["version"]:
> + col = {}
> + col["version"] = row["version"]
> + col["pkgarch"] = row["pkgarch"]
> + col["checksum"] = row["checksum"]
> + col["value"] = row["value"]
> + datainfo.append(col)
> return (metainfo, datainfo)
>
> def dump_db(self, fd):
> @@ -345,14 +323,15 @@ class PRData(object):
> raise e
> uri = "file:%s%s" % (self.filename, "?mode=ro" if self.read_only
> else "") logger.debug("Opening PRServ database '%s'" % (uri))
> - self.connection=sqlite3.connect(uri, uri=True,
> isolation_level="EXCLUSIVE", check_same_thread = False) +
> self.connection=sqlite3.connect(uri, uri=True)
> self.connection.row_factory=sqlite3.Row
> - if not self.read_only:
> - self.connection.execute("pragma synchronous = off;")
> - self.connection.execute("PRAGMA journal_mode = MEMORY;")
> + self.connection.execute("PRAGMA synchronous = OFF;")
> + self.connection.execute("PRAGMA journal_mode = WAL;")
> + self.connection.commit()
> self._tables={}
>
> def disconnect(self):
> + self.connection.commit()
> self.connection.close()
>
> def __getitem__(self, tblname):
> @@ -370,3 +349,4 @@ class PRData(object):
> del self._tables[tblname]
> logger.info("drop table %s" % (tblname))
> self.connection.execute("DROP TABLE IF EXISTS %s;" % tblname)
> + self.connection.commit()
> diff --git a/lib/prserv/serv.py b/lib/prserv/serv.py
> index 05573d06cc..fd673b1851 100644
> --- a/lib/prserv/serv.py
> +++ b/lib/prserv/serv.py
> @@ -44,8 +44,6 @@ class PRServerClient(bb.asyncrpc.AsyncServerConnection):
> except:
> self.server.table.sync()
> raise
> - else:
> - self.server.table.sync_if_dirty()
>
> async def handle_test_pr(self, request):
> '''Finds the PR value corresponding to the request. If not found,
> returns None and doesn't insert a new value''' @@ -233,15 +231,9 @@ class
> PRServer(bb.asyncrpc.AsyncServer):
> return tasks
>
> async def stop(self):
> - self.table.sync_if_dirty()
> self.db.disconnect()
> await super().stop()
>
> - def signal_handler(self):
> - super().signal_handler()
> - if self.table:
> - self.table.sync()
> -
> class PRServSingleton(object):
> def __init__(self, dbfile, logfile, host, port, upstream):
> self.dbfile = dbfile
^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [bitbake-devel] [PATCH v6 0/8] prserv: add support for an "upstream" server
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
` (7 preceding siblings ...)
2024-04-30 17:15 ` [PATCH v6 8/8] prserv: add bitbake selftests michael.opdenacker
@ 2024-04-30 20:31 ` Jan-Simon Moeller
8 siblings, 0 replies; 11+ messages in thread
From: Jan-Simon Moeller @ 2024-04-30 20:31 UTC (permalink / raw)
To: michael.opdenacker; +Cc: bitbake-devel
Reviewed-by: Jan-Simon Möller <dl9pf@gmx.de>
Am Dienstag, 30. April 2024, 19:15:04 CEST schrieb Michael Opdenacker via
lists.openembedded.org:
> From: Michael Opdenacker <michael.opdenacker@bootlin.com>
>
> This makes it possible to customize an "upstream" distribution
> by modifying local packages. If the "upstream" package bears
> revision "x", the local one will have revision "x.y", this
> having priority over the upstream one.
>
> Multiple levels of upstream servers are supported, so "x.y.z" revisions
> are possible too.
>
> This version also supports multiple servers sharing the same database.
>
> This also includes BitBake selftests to check the expected
> functionality and detect future regressions.
>
> Note that database import and export functions aren't tested yet.
>
> ---
>
> Changes in V6:
>
> - Harden the store_value() code to protect
> it against storing the same value twice.
>
> - Simplify the importone() hook, removing
> its unnecessary "history" parameter.
>
> - This fixes all errors in "oe-selftest -r prservice",
> once this OE-core patch is merged:
>
> https://lore.kernel.org/openembedded-core/20240429194656.655509-1-michael.o
> pdenacker@bootlin.com/T/#u
>
> - Make tests work in temporary directories.
> Suggested by Joshua Watt
>
> - Remove first version of readonly server tests
> which depended on the output of a previous test.
> One implementation of readonly server tests remains.
>
> - Merge the upstream and read-only tests so that
> the read-only test part can reuse the databases
> and servers from the upstream one. Initially thought
> that the setUp (and tearDown) hooks were called only
> once for each test class, instead of for each test.
> Thanks to Joshua Watt for making me understand this.
>
> - Merged two more tests and simplified their naming
> as, with one exception, the order of tests doesn't
> matter any more.
>
> - Avoid a possible race condition at database
> creation time. Suggested by Joshua Watt.
>
> Changes in V5:
>
> - Change the way the database is accessed and modified.
> As done in the Hash Server code, the database is not
> accessed through the cursor API.
>
> This makes it possible to have multiple servers accessing
> the same database, typically one read-write server that
> is fed by the build machine, and one read-only server that
> is used to publish PR data to downstream users.
>
> This switched sqlite3 to "WAL" journaling mode, and to
> "OFF" synchronous mode.
>
> - Fix revision ordering issues, now computing the min and max
> values through dedicated Python functions, instead of letting
> the database do this with just string comparison, which
> was wrong in our case. Typically, "1.20" is greater for us
> than "1.3".
>
> - Add BitBake selftests, in particular to make sure revision
> ordering is correct and the maximum and minimum values are
> correctly computed.
>
> - An issue remains: the BitBake selftests that test database
> access through a read-only server while a read-write server
> is still running, still fail, because the database changes
> are apparently not stored to disk.
>
> FAIL: test_3b_readonly (prserv.tests.PRUpstreamTests)
>
> Changes in V4:
>
> - Add BitBake selftests for the legacy and new PR server features
> (database, client, server, support for upstream server, read-only mode,
> "history" and "no history" modes) To run only these tests:
> bitbake-selftest prserv.tests
>
> - Pass the "history" mode through the client requests, instead
> of storing it (globally) in the database name.
>
> The PR database is now called "PRMAIN" instead of "PRMAIN_nohist".
> This should cause a regression for builds which already
> have a PR database.
>
> - Fixes for "history" modes:
> - Allow to store multiple PR values for the same checksum,
> needed for the "no history" mode.
> - Make the "history" mode return the minimum stored
> PR value.
>
> - Fixes and code reorganization for issues uncovered by the tests.
>
> - Update the server version to "2.0.0"
>
> Changes in V3:
>
> - Revert the commit removing the so far unused "hist" mode, which
> we wish to keep for binary reproducibility sake.
>
> - Simplification of get_value() function to take
> both "hist" and "nohist" modes with the same shared code.
>
> - Add "history" parameter to the "getPR" request,
> so that the client can ask for the mode of its choice.
> This will also make it possible to implement tests
> for both modes.
>
> Note that more requests ("export", "import"...)
> will also need a "history" parameter, in a future version,
> after the first tests are implemented.
>
> - Several bug fixes.
>
> - Put all the new features at the tip of the branch,
> to make the cleanup commits easier to merged.
>
> Changes in V2:
>
> - Add this new commit:
> prserv: remove unused "hist" mode in the database backend
>
> - Squash commit "prserv: fix read_only test" into
> commit "prserv: simplify the PRServerClient() interface"
> (Reported by Richard Purdie)
>
> - Fix the code to support increasing "x.y.z" values, thus
> supporting several levels of upstream servers.
>
> - db.py: remove duplicate definition of find_max_value() function in db.py
>
> - prserv.py: remove tabs before comments (Python didn't complain)
>
> - db.py: now stores the revision ("value") as TEXT.
> This way we can store "1.0" without having it transformed to "1"
> when the default type was INTEGER.
>
> - This allows to fix a regression when the first packages were created
> with 'r0.1' instead of 'r0.0' initially.
>
> - find_max_value: now returns None instead of '0' when no value is found
> Before we couldn't tell the difference between a '0'
> max value and the absence of such a value.
>
> Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
> Cc: Joshua Watt <JPEWhacker@gmail.com>
> Cc: Tim Orling <ticotimo@gmail.com>
>
> Michael Opdenacker (8):
> prserv: declare "max_package_pr" client hook
> prserv: move code from __init__ to bitbake-prserv
> prserv: add "upstream" server support
> prserv: enable database sharing
> prserv: avoid possible race condition in database code
> prserv: store_value() improvements
> prserv: import simplification
> prserv: add bitbake selftests
>
> bin/bitbake-prserv | 26 ++-
> bin/bitbake-selftest | 2 +
> lib/prserv/__init__.py | 97 ++++++++-
> lib/prserv/client.py | 15 +-
> lib/prserv/db.py | 452 ++++++++++++++++++-----------------------
> lib/prserv/serv.py | 139 ++++++++++---
> lib/prserv/tests.py | 386 +++++++++++++++++++++++++++++++++++
> 7 files changed, 816 insertions(+), 301 deletions(-)
> create mode 100644 bitbake/lib/prserv/tests.py
^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2024-04-30 20:31 UTC | newest]
Thread overview: 11+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2024-04-30 17:15 [PATCH v6 0/8] prserv: add support for an "upstream" server michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 1/8] prserv: declare "max_package_pr" client hook michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 2/8] prserv: move code from __init__ to bitbake-prserv michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 3/8] prserv: add "upstream" server support michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 4/8] prserv: enable database sharing michael.opdenacker
2024-04-30 20:29 ` [bitbake-devel] " Jan-Simon Moeller
2024-04-30 17:15 ` [PATCH v6 5/8] prserv: avoid possible race condition in database code michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 6/8] prserv: store_value() improvements michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 7/8] prserv: import simplification michael.opdenacker
2024-04-30 17:15 ` [PATCH v6 8/8] prserv: add bitbake selftests michael.opdenacker
2024-04-30 20:31 ` [bitbake-devel] [PATCH v6 0/8] prserv: add support for an "upstream" server Jan-Simon Moeller
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.