All of lore.kernel.org
 help / color / mirror / Atom feed
From: "Yan, Zheng" <zheng.z.yan@intel.com>
To: Gregory Farnum <greg@inktank.com>
Cc: "ceph-devel@vger.kernel.org" <ceph-devel@vger.kernel.org>,
	Sage Weil <sage@inktank.com>
Subject: Re: [PATCH 13/39] mds: don't send resolve message between active MDS
Date: Thu, 21 Mar 2013 10:55:41 +0800	[thread overview]
Message-ID: <514A76AD.5030005@intel.com> (raw)
In-Reply-To: <CAPYLRzj0BYniL2g21qhFFV0OvhZU-O+btnTFAfsj7=qPR-b_mw@mail.gmail.com>

On 03/21/2013 05:56 AM, Gregory Farnum wrote:
> On Sun, Mar 17, 2013 at 7:51 AM, Yan, Zheng <zheng.z.yan@intel.com> wrote:
>> From: "Yan, Zheng" <zheng.z.yan@intel.com>
>>
>> When MDS cluster is resolving, current behavior is sending subtree resolve
>> message to all other MDS and waiting for all other MDS' resolve message.
>> The problem is that active MDS can have diffent subtree map due to rename.
>> Besides gathering active MDS's resolve messages are also racy. The only
>> function for these messages is disambiguate other MDS' import. We can
>> replace it by import finish notification.
>>
>> Signed-off-by: Yan, Zheng <zheng.z.yan@intel.com>
>> ---
>>  src/mds/MDCache.cc  | 12 +++++++++---
>>  src/mds/Migrator.cc | 25 +++++++++++++++++++++++--
>>  src/mds/Migrator.h  |  3 ++-
>>  3 files changed, 34 insertions(+), 6 deletions(-)
>>
>> diff --git a/src/mds/MDCache.cc b/src/mds/MDCache.cc
>> index c455a20..73c1d59 100644
>> --- a/src/mds/MDCache.cc
>> +++ b/src/mds/MDCache.cc
>> @@ -2517,7 +2517,8 @@ void MDCache::send_subtree_resolves()
>>         ++p) {
>>      if (*p == mds->whoami)
>>        continue;
>> -    resolves[*p] = new MMDSResolve;
>> +    if (mds->is_resolve() || mds->mdsmap->is_resolve(*p))
>> +      resolves[*p] = new MMDSResolve;
>>    }
>>
>>    // known
>> @@ -2837,7 +2838,7 @@ void MDCache::handle_resolve(MMDSResolve *m)
>>           migrator->import_reverse(dir);
>>         } else {
>>           dout(7) << "ambiguous import succeeded on " << *dir << dendl;
>> -         migrator->import_finish(dir);
>> +         migrator->import_finish(dir, true);
>>         }
>>         my_ambiguous_imports.erase(p);  // no longer ambiguous.
>>        }
>> @@ -3432,7 +3433,12 @@ void MDCache::rejoin_send_rejoins()
>>         ++p) {
>>      CDir *dir = p->first;
>>      assert(dir->is_subtree_root());
>> -    assert(!dir->is_ambiguous_dir_auth());
>> +    if (dir->is_ambiguous_dir_auth()) {
>> +      // exporter is recovering, importer is survivor.
> 
> The importer has to be the MDS this code is running on, right?

This code is for bystanders. The exporter is recovering, and its resolve message didn't claim
the subtree. So the export must succeed.

> 
>> +      assert(rejoins.count(dir->authority().first));
>> +      assert(!rejoins.count(dir->authority().second));
>> +      continue;
>> +    }
>>
>>      // my subtree?
>>      if (dir->is_auth())
>> diff --git a/src/mds/Migrator.cc b/src/mds/Migrator.cc
>> index 5e53803..833df12 100644
>> --- a/src/mds/Migrator.cc
>> +++ b/src/mds/Migrator.cc
>> @@ -2088,6 +2088,23 @@ void Migrator::import_reverse(CDir *dir)
>>    }
>>  }
>>
>> +void Migrator::import_notify_finish(CDir *dir, set<CDir*>& bounds)
>> +{
>> +  dout(7) << "import_notify_finish " << *dir << dendl;
>> +
>> +  for (set<int>::iterator p = import_bystanders[dir].begin();
>> +       p != import_bystanders[dir].end();
>> +       ++p) {
>> +    MExportDirNotify *notify =
>> +      new MExportDirNotify(dir->dirfrag(), false,
>> +                          pair<int,int>(import_peer[dir->dirfrag()], mds->get_nodeid()),
>> +                          pair<int,int>(mds->get_nodeid(), CDIR_AUTH_UNKNOWN));
> 
> I don't think this is quite right — we're notifying them that we've
> just finished importing data from somebody, right? And so we know that
> we're the auth node...

Yes. In normal case, exporter notifies the bystanders. But if exporter crashes, the importer notifies
the bystanders after it confirms ambiguous import succeeds.

Thanks
Yan, Zheng

> 
>> +    for (set<CDir*>::iterator i = bounds.begin(); i != bounds.end(); i++)
>> +      notify->get_bounds().push_back((*i)->dirfrag());
>> +    mds->send_message_mds(notify, *p);
>> +  }
>> +}
>> +
>>  void Migrator::import_notify_abort(CDir *dir, set<CDir*>& bounds)
>>  {
>>    dout(7) << "import_notify_abort " << *dir << dendl;
>> @@ -2183,11 +2200,11 @@ void Migrator::handle_export_finish(MExportDirFinish *m)
>>    CDir *dir = cache->get_dirfrag(m->get_dirfrag());
>>    assert(dir);
>>    dout(7) << "handle_export_finish on " << *dir << dendl;
>> -  import_finish(dir);
>> +  import_finish(dir, false);
>>    m->put();
>>  }
>>
>> -void Migrator::import_finish(CDir *dir)
>> +void Migrator::import_finish(CDir *dir, bool notify)
>>  {
>>    dout(7) << "import_finish on " << *dir << dendl;
>>
>> @@ -2205,6 +2222,10 @@ void Migrator::import_finish(CDir *dir)
>>    // remove pins
>>    set<CDir*> bounds;
>>    cache->get_subtree_bounds(dir, bounds);
>> +
>> +  if (notify)
>> +    import_notify_finish(dir, bounds);
>> +
>>    import_remove_pins(dir, bounds);
>>
>>    map<CInode*, map<client_t,Capability::Export> > cap_imports;
>> diff --git a/src/mds/Migrator.h b/src/mds/Migrator.h
>> index 7988f32..2889a74 100644
>> --- a/src/mds/Migrator.h
>> +++ b/src/mds/Migrator.h
>> @@ -273,12 +273,13 @@ protected:
>>    void import_reverse_unfreeze(CDir *dir);
>>    void import_reverse_final(CDir *dir);
>>    void import_notify_abort(CDir *dir, set<CDir*>& bounds);
>> +  void import_notify_finish(CDir *dir, set<CDir*>& bounds);
>>    void import_logged_start(dirfrag_t df, CDir *dir, int from,
>>                            map<client_t,entity_inst_t> &imported_client_map,
>>                            map<client_t,uint64_t>& sseqmap);
>>    void handle_export_finish(MExportDirFinish *m);
>>  public:
>> -  void import_finish(CDir *dir);
>> +  void import_finish(CDir *dir, bool notify);
>>  protected:
>>
>>    void handle_export_caps(MExportCaps *m);
>> --
>> 1.7.11.7
>>

--
To unsubscribe from this list: send the line "unsubscribe ceph-devel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

  reply	other threads:[~2013-03-21  2:55 UTC|newest]

Thread overview: 117+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-03-17 14:51 [PATCH 00/39] fixes for MDS cluster recovery Yan, Zheng
2013-03-17 14:51 ` [PATCH 01/39] mds: preserve subtree bounds until slave commit Yan, Zheng
2013-03-20 18:33   ` Greg Farnum
2013-03-17 14:51 ` [PATCH 02/39] mds: process finished contexts in batch Yan, Zheng
2013-03-20 18:33   ` Greg Farnum
2013-03-17 14:51 ` [PATCH 03/39] mds: fix MDCache::adjust_bounded_subtree_auth() Yan, Zheng
2013-03-20 18:33   ` Greg Farnum
2013-03-17 14:51 ` [PATCH 04/39] mds: make sure table request id unique Yan, Zheng
2013-03-19 23:09   ` Greg Farnum
2013-03-20  5:53     ` Yan, Zheng
2013-03-20  6:15       ` Sage Weil
2013-03-20  6:24         ` Yan, Zheng
2013-03-20  6:49         ` Yan, Zheng
2013-03-20 18:31           ` Greg Farnum
2013-03-21  8:07             ` Yan, Zheng
2013-03-21 22:03               ` Gregory Farnum
2013-03-25 11:30                 ` Yan, Zheng
2013-03-29 22:12                   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 05/39] mds: send table request when peer is in proper state Yan, Zheng
2013-03-20 18:34   ` Greg Farnum
2013-03-29 21:58   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 06/39] mds: make table client/server tolerate duplicated message Yan, Zheng
2013-03-29 22:00   ` Gregory Farnum
2013-03-31 13:21     ` Yan, Zheng
2013-03-17 14:51 ` [PATCH 07/39] mds: mark connection down when MDS fails Yan, Zheng
2013-03-20 18:37   ` Greg Farnum
2013-03-17 14:51 ` [PATCH 08/39] mds: consider MDS as recovered when it reaches clientreply state Yan, Zheng
2013-03-20 18:40   ` Greg Farnum
2013-03-21  2:22     ` Yan, Zheng
2013-03-21 21:43       ` Gregory Farnum
2013-03-20 19:09   ` Greg Farnum
2013-03-17 14:51 ` [PATCH 09/39] mds: defer eval gather locks when removing replica Yan, Zheng
2013-03-20 19:36   ` Greg Farnum
2013-03-21  2:29     ` Yan, Zheng
2013-03-17 14:51 ` [PATCH 10/39] mds: unify slave request waiting Yan, Zheng
2013-03-20 22:52   ` Sage Weil
2013-03-17 14:51 ` [PATCH 11/39] mds: don't delay processing replica buffer in slave request Yan, Zheng
2013-03-20 21:19   ` Greg Farnum
2013-03-21  2:38     ` Yan, Zheng
2013-03-21  4:15       ` Sage Weil
2013-03-21 21:48         ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 12/39] mds: compose and send resolve messages in batch Yan, Zheng
2013-03-20 21:45   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 13/39] mds: don't send resolve message between active MDS Yan, Zheng
2013-03-20 21:56   ` Gregory Farnum
2013-03-21  2:55     ` Yan, Zheng [this message]
2013-03-21 21:55       ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 14/39] mds: set resolve/rejoin gather MDS set in advance Yan, Zheng
2013-03-20 22:09   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 15/39] mds: don't send MDentry{Link,Unlink} before receiving cache rejoin Yan, Zheng
2013-03-20 22:17   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 16/39] mds: send cache rejoin messages after gathering all resolves Yan, Zheng
2013-03-20 22:57   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 17/39] mds: send resolve acks after master updates are safely logged Yan, Zheng
2013-03-20 22:58   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 18/39] mds: fix MDS recovery involving cross authority rename Yan, Zheng
2013-03-21 17:59   ` Gregory Farnum
2013-03-22  3:04     ` Yan, Zheng
2013-03-29 22:02       ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 19/39] mds: remove MDCache::rejoin_fetch_dirfrags() Yan, Zheng
2013-03-20 22:58   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 20/39] mds: include replica nonce in MMDSCacheRejoin::inode_strong Yan, Zheng
2013-03-20 23:26   ` Gregory Farnum
2013-03-20 23:36     ` Sage Weil
2013-03-17 14:51 ` [PATCH 21/39] mds: encode dirfrag base in cache rejoin ack Yan, Zheng
2013-03-20 23:33   ` Gregory Farnum
2013-03-20 23:40     ` Gregory Farnum
2013-03-21  6:41     ` Yan, Zheng
2013-03-21 21:58       ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 22/39] mds: handle linkage mismatch during cache rejoin Yan, Zheng
2013-03-21 21:23   ` Gregory Farnum
2013-03-22  3:05     ` Yan, Zheng
2013-03-25 16:14       ` Gregory Farnum
2013-03-26  7:21     ` Yan, Zheng
2013-03-29 22:09       ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 23/39] mds: reqid for rejoinning authpin/wrlock need to be list Yan, Zheng
2013-03-20 23:59   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 24/39] mds: take object's versionlock when rejoinning xlock Yan, Zheng
2013-03-21  0:37   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 25/39] mds: share inode max size after MDS recovers Yan, Zheng
2013-03-21  0:45   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 26/39] mds: issue caps when lock state in replica become SYNC Yan, Zheng
2013-03-21  0:52   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 27/39] mds: send lock action message when auth MDS is in proper state Yan, Zheng
2013-03-21  3:12   ` Gregory Farnum
2013-03-21  3:20     ` Yan, Zheng
2013-03-17 14:51 ` [PATCH 28/39] mds: add dirty imported dirfrag to LogSegment Yan, Zheng
2013-03-21  3:14   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 29/39] mds: avoid double auth pin for file recovery Yan, Zheng
2013-03-21  3:20   ` Gregory Farnum
2013-03-21  3:33     ` Yan, Zheng
2013-03-21  4:20       ` Sage Weil
2013-03-21 21:58     ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 30/39] mds: check MDS peer's state through mdsmap Yan, Zheng
2013-03-21  3:24   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 31/39] mds: unfreeze subtree if import aborts in PREPPED state Yan, Zheng
2013-03-21  3:27   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 32/39] mds: fix export cancel notification Yan, Zheng
2013-03-21  3:31   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 33/39] mds: notify bystanders if export aborts Yan, Zheng
2013-03-21  3:34   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 34/39] mds: don't open dirfrag while subtree is frozen Yan, Zheng
2013-03-21  3:38   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 35/39] mds: clear dirty inode rstat if import fails Yan, Zheng
2013-03-21  3:40   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 36/39] mds: try merging subtree after clear EXPORTBOUND Yan, Zheng
2013-03-21  3:44   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 37/39] mds: eval inodes with caps imported by cache rejoin message Yan, Zheng
2013-03-21  3:45   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 38/39] mds: don't replicate purging dentry Yan, Zheng
2013-03-21  3:46   ` Gregory Farnum
2013-03-17 14:51 ` [PATCH 39/39] mds: clear scatter dirty if replica inode has no auth subtree Yan, Zheng
2013-03-21  3:49   ` Gregory Farnum
2013-04-01  8:46 ` [PATCH 00/39] fixes for MDS cluster recovery Yan, Zheng
2013-04-01 17:00   ` Gregory Farnum
2013-04-01  8:51 ` [PATCH] mds: avoid sending duplicated table prepare/commit Yan, Zheng
2013-04-01  8:51   ` [PATCH] mds: don't roll back prepared table updates Yan, Zheng

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=514A76AD.5030005@intel.com \
    --to=zheng.z.yan@intel.com \
    --cc=ceph-devel@vger.kernel.org \
    --cc=greg@inktank.com \
    --cc=sage@inktank.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 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.