Bug 1224292 - peers connected in the middle of a transaction are participating in the transaction
Summary: peers connected in the middle of a transaction are participating in the trans...
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: GlusterFS
Classification: Community
Component: glusterd
Version: 3.7.0
Hardware: Unspecified
OS: Unspecified
unspecified
unspecified
Target Milestone: ---
Assignee: Avra Sengupta
QA Contact:
URL:
Whiteboard:
Depends On: 1224290
Blocks: glusterfs-3.7.1 1225371
TreeView+ depends on / blocked
 
Reported: 2015-05-22 13:05 UTC by Avra Sengupta
Modified: 2015-12-01 16:45 UTC (History)
3 users (show)

Fixed In Version: glusterfs-3.7.1
Clone Of: 1224290
: 1225371 (view as bug list)
Environment:
Last Closed: 2015-06-02 08:03:11 UTC
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Embargoed:


Attachments (Terms of Use)

Description Avra Sengupta 2015-05-22 13:05:17 UTC
+++ This bug was initially created as a clone of Bug #1224290 +++

Description of problem:
peers which are probed and awaiting connection, when get connected in between a transaction, cause unexpected behaviour with the operation.

Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

Comment 1 Anand Avati 2015-05-27 09:29:31 UTC
REVIEW: http://review.gluster.org/10937 (glusterd: Fix conf->generation to stop new peers participating in a transaction, while the transaction is in progress.) posted (#1) for review on release-3.7 by Avra Sengupta (asengupt)

Comment 2 Anand Avati 2015-05-27 15:38:56 UTC
COMMIT: http://review.gluster.org/10937 committed in release-3.7 by Vijay Bellur (vbellur) 
------
commit 0bcd638b46a7134f0285b39aabfed9950645bdc3
Author: Avra Sengupta <asengupt>
Date:   Fri May 22 18:38:23 2015 +0530

    glusterd: Fix conf->generation to stop new peers participating in
    a transaction, while the transaction is in progress.
    
            Backport of http://review.gluster.org/#/c/10895/
    
    Every peer gets a generation number, during it's inception.
    This generation number is used to identify the peer throiughout
    it's lifetime. This number is assigned based on the current
    generation number of the system, which is incremented with every
    peer that is added.
    
    The problem arises when we add a peer, and before it gets a rpc_connect
    we begin a transaction. In such a case, the peer gets considered in the
    transaction, but doesn't participate in it coz it isn't connected yet.
    The moment it gets the rpc notification and is connected, it starts
    participating in the transaction and all hell breaks loose.
    
    To resolve it, we should assign the peerinfo a new generation number
    everytime it's connected, so that this number will be greater than the
    generation number that the transaction is acting upon, and even though
    the peer is connected it will not participate in the transaction.
    
    We should also assign the new generation number of the peer to the peerctx,
    so that the framework that searches for peerinfos based on the generation
    number, will still function in the same manner.
    
    Removing ./tests/basic/volume-snapshot-clone.t from bad-tests.
    Also removed the duplicate entry of ./tests/bugs/snapshot/bug-1112559.t
    from bad-tests. Original entry was removed in
    http://review.gluster.org/10840
    
    Change-Id: I57e5efe3217abfb96733e487aea3972e116d909d
    BUG: 1224292
    Signed-off-by: Avra Sengupta <asengupt>
    Reviewed-on: http://review.gluster.org/10895
    Reviewed-by: Kaushal M <kaushal>
    Tested-by: Kaushal M <kaushal>
    Reviewed-on: http://review.gluster.org/10937
    Reviewed-by: Atin Mukherjee <amukherj>
    Tested-by: Gluster Build System <jenkins.com>
    Tested-by: NetBSD Build System
    Reviewed-by: Vijay Bellur <vbellur>

Comment 3 Niels de Vos 2015-06-02 08:03:11 UTC
This bug is getting closed because a release has been made available that should address the reported issue. In case the problem is still not fixed with glusterfs-3.7.1, please reopen this bug report.

glusterfs-3.7.1 has been announced on the Gluster Packaging mailinglist [1], packages for several distributions should become available in the near future. Keep an eye on the Gluster Users mailinglist [2] and the update infrastructure for your distribution.

[1] http://thread.gmane.org/gmane.comp.file-systems.gluster.packaging/1
[2] http://thread.gmane.org/gmane.comp.file-systems.gluster.user


Note You need to log in before you can comment on or make changes to this bug.