From: Eryu Guan <eguan@linux.alibaba.com>
To: Ira Weiny <ira.weiny@intel.com>
Cc: Eryu Guan <guaneryu@gmail.com>,
fstests@vger.kernel.org, john.hubbard@gmail.com,
Dave Chinner <david@fromorbit.com>,
Jason Gunthorpe <jgg@ziepe.ca>, Jan Kara <jack@suse.cz>,
Dan Williams <dan.j.williams@intel.com>,
Jeff Layton <jlayton@kernel.org>
Subject: Re: [PATCH V2 2/2] generic: Add Lease testing
Date: Tue, 10 Sep 2019 10:29:55 +0800 [thread overview]
Message-ID: <20190910022955.GY52397@e18g06458.et15sqa> (raw)
In-Reply-To: <20190909153328.GA11186@iweiny-DESK2.sc.intel.com>
On Mon, Sep 09, 2019 at 08:33:29AM -0700, Ira Weiny wrote:
> On Sun, Sep 08, 2019 at 10:33:12PM +0800, Eryu Guan wrote:
> > On Thu, Sep 05, 2019 at 10:30:26AM -0700, ira.weiny@intel.com wrote:
> > > From: Ira Weiny <ira.weiny@intel.com>
> > >
> > > Now that we have a leasetest executable add in a lease test script.
> > >
> > > Signed-off-by: Ira Weiny <ira.weiny@intel.com>
> >
> > The new test could be merged with patch 1.
>
> Ok, I was not sure how best to submit this. Now I know.
> >
> > > ---
> > > tests/generic/567 | 100 ++++++++++++++++++++++++++++++++++++++++++
> > > tests/generic/567.out | 2 +
> > > tests/generic/group | 1 +
> > > 3 files changed, 103 insertions(+)
> > > create mode 100755 tests/generic/567
> > > create mode 100644 tests/generic/567.out
> > >
> > > diff --git a/tests/generic/567 b/tests/generic/567
> > > new file mode 100755
> > > index 000000000000..3ef99c66b207
> > > --- /dev/null
> > > +++ b/tests/generic/567
> > > @@ -0,0 +1,100 @@
> > > +#! /bin/bash
> > > +# SPDX-License-Identifier: GPL-2.0
> > > +# Copyright (c) 2019 Intel Corp. All Rights Reserved.
> > > +#
> > > +# FS QA Test 567
> > > +#
> > > +# Test Leases
> > > +#
> > > +seq=`basename $0`
> > > +seqres=$RESULT_DIR/$seq
> > > +echo "QA output created by $seq"
> > > +
> > > +here=`pwd`
> > > +tmp=/tmp/$$
> > > +status=1 # failure is the default!
> > > +trap "_cleanup; exit \$status" 0 1 2 3 15
> > > +#debug="-d -d -d"
> > > +debug=""
> >
> > No need to define 'debug', if you need debug output, make 'debug' the
> > default and dump all debug output to $seqres.full file.
>
> I'll check this out. I'm new to the harness so I was not sure what got output
> where.
Basically, all stdout and stderr outputs from the test script are
recorded by the harness and compared with the golden output (the .out
file) after test, and fail the test if they are different. And
$seqres.full is a place for holding verbose test outputs for debug
purpose, which are used to help determine what's really going wrong.
>
> FWIW I think that means we can simplify leasetest even more then. Remove all
> the "debug" flags and leave stderr for error condition details.
Only if we dump all the outputs to $seqres.full and test only depends on
the return value of leasetest.
>
> Anyway, your comment WRT leasetest is that you would like to see locktest
> "modernized" to include lease testing? Is that correct? If so, would you like
> me to make these script modifications to 131? Or would you like to see 2
> "modes" of locktest?
>
> I'm not convinced that overloading the parameters of the test array in locktest
> is a good idea. There are already a number of confusing overloads there. And
> leasetest is going to get some of its own which will further complicate things.
> (I have added truncate tests to lease testing as well.)
I was thinking of two modes, not overloading the parameters of single
test array, i.e. define two test arrays, one for locktest and another
for lease test, and using which array is determined by command switch.
But either way is fine if we could reuse the framework from locktest.
>
> >
> > > +
> > > +TESTFILE=$TEST_DIR/lease_file
> > > +
> > > +_cleanup()
> > > +{
> > > + kill $leasetest_pid2 > /dev/null 2>&1
> > > + kill $leasetest_pid1 > /dev/null 2>&1
> > > + rm -f $TESTFILE
> > > +}
> > > +
> > > +# get standard environment, filters and checks
> > > +. ./common/rc
> > > +. ./common/filter
> > > +
> > > +# real QA test starts here
> > > +_supported_fs generic
> > > +_supported_os Linux
> > > +_require_test
> > > +_require_test_fcntl_advisory_locks
> > > +
> > > +# set up log files
> > > +SERVER_LOG=$TEST_DIR/server.out
> > > +CLIENT_LOG=$TEST_DIR/client.out
> > > +rm -f $SERVER_LOG
> > > +touch $SERVER_LOG
> > > +rm -f $CLIENT_LOG
> > > +touch $CLIENT_LOG
> > > +
> > > +if [ "$debug" != "" ]; then
> > > + echo "Test directory : $TEST_DIR"
> > > + echo "Result directory : $RESULT_DIR"
> > > + echo "Client log : $CLIENT_LOG"
> > > + echo "Server log : $SERVER_LOG"
> > > +fi
> > > +
> > > +touch $TESTFILE
> > > +
> > > +# Start the server
> > > +src/leasetest $debug $TESTFILE 2>&1 > $SERVER_LOG &
> > > +leasetest_pid1=$!
> > > +
> > > +timeout=30
> > > +while [ $timeout -gt 0 ]; do
> > > + sleep 1
> > > +
> > > + PORT=$(cat $SERVER_LOG | grep "^server port: " | awk '{print $3}')
> > > + if [ -n "$PORT" ]; then
> > > + break
> > > + fi
> > > +
> > > + # check the existence of server process
> > > + if ! kill -s 0 $leasetest_pid1 >/dev/null 2>&1; then
> > > + echo "Server died abnormally"
> > > + exit 1
> > > + fi
> > > +
> > > + let timeout=timeout-1
> > > +done
> > > +
> > > +if [ -z "$PORT" ]; then
> > > + echo "Could not get server port"
> > > + exit 1
> > > +fi
> >
> > The whole "discovering server port" part in generic/131 has been patched
> > several times in the past to fix corner cases here and there, we'd
> > better to refactor this hunk into a common helper, instead of
> > duplicating the code.
>
> You mean refactor this to be common between this and the generic/131
> (locktest)?
Correct.
>
> But see my comment above.
>
> >
> > > +
> > > +# Start the client
> > > +src/leasetest $debug -p $PORT -h localhost $TESTFILE 2>&1 > $CLIENT_LOG
> > > +leasetest_pid2=$!
> > > +result=$?
> > > +if [ $result -eq 0 ]; then
> > > + echo success!
> > > + status=0
> > > +else
> > > + echo "Client reported failure ($result)"
> > > + cat $TEST_DIR/*.out
> > > +fi
> > > +
> > > +if [ "$debug" != "" ]; then
> > > + echo "End file details"
> > > + ls -la $TESTFILE
> > > +fi
> > > +
> > > +exit
> >
> > echo "Silence is golden"
> > src/leasetest ...
> > leasetest_pid2=$!
> > result=$?
> > if [ $result -ne 0 ]; then
> > echo "Server log" >> $seqres.full
> > cat $SERVER_LOG >> $seqres.full
> > echo "Client log" >> $seqres.full
> > cat $CLIENT_LOG >> $seqres.full
> > _fail "Client reported failure ($result)"
> > fi
> > status=0
> > exit
> >
> > _fail will output the failure message, which breaks the golden output,
> > and prompt user to check $seqres.full file for details.
>
> Ah ok I see. Sorry I was not aware of $seqres.full My bad...
Not a problem :)
>
> I'll see if I can clean up what is really "debugging" vs not as well within
> leasetest.
Thanks for the test!
Eryu
prev parent reply other threads:[~2019-09-10 2:30 UTC|newest]
Thread overview: 7+ messages / expand[flat|nested] mbox.gz Atom feed top
2019-09-05 17:30 [PATCH V2 0/2] Add lease testing ira.weiny
2019-09-05 17:30 ` [PATCH V2 1/2] src/leasetest: Add lease test executable ira.weiny
2019-09-08 14:17 ` Eryu Guan
2019-09-05 17:30 ` [PATCH V2 2/2] generic: Add Lease testing ira.weiny
2019-09-08 14:33 ` Eryu Guan
2019-09-09 15:33 ` Ira Weiny
2019-09-10 2:29 ` Eryu Guan [this message]
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20190910022955.GY52397@e18g06458.et15sqa \
--to=eguan@linux.alibaba.com \
--cc=dan.j.williams@intel.com \
--cc=david@fromorbit.com \
--cc=fstests@vger.kernel.org \
--cc=guaneryu@gmail.com \
--cc=ira.weiny@intel.com \
--cc=jack@suse.cz \
--cc=jgg@ziepe.ca \
--cc=jlayton@kernel.org \
--cc=john.hubbard@gmail.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox