[Gluster-devel] Rebalance is not working in single node cluster environment.

[Anand Nekkunti]

On 06/13/2015 04:50 PM, Atin Mukherjee wrote:
>
 > Sent from Samsung Galaxy S4 On 13 Jun 2015 14:42, "Anand Nekkunti"
 > <anekkunt at redhat.com <mailto:anekkunt at redhat.com>> wrote:
 >>
 >>
 >> On 06/13/2015 02:27 PM, Atin Mukherjee wrote:
 >>>
 >>> Sent from Samsung Galaxy S4 On 13 Jun 2015 13:15, "Raghavendra
 >>> Talur" <raghavendra.talur at gmail.com
 >>> <mailto:raghavendra.talur at gmail.com>> wrote:
 >>>>
 >>>>
 >>>>
 >>>> On Sat, Jun 13, 2015 at 1:00 PM, Atin Mukherjee
 >>>> <atin.mukherjee83 at gmail.com
 >>>> <mailto:atin.mukherjee83 at gmail.com>> wrote:
 >>>>>
 >>>>> Sent from Samsung Galaxy S4 On 13 Jun 2015 12:58, "Anand
 >>>>> Nekkunti" <anekkunt at redhat.com <mailto:anekkunt at redhat.com>>
 >>>>> wrote:
 >>>>>>
 >>>>>> Hi All Rebalance is not working in single node cluster
 >>>>>> environment ( current test frame work ). I am getting
 >>>>>> error in below test , it seems re-balance is not migrated
 >>>>>> to  current cluster test framework.
 >>>>> Could you pin point which test case fails and what log do you
 >>>>> see?
 >>>>>>
 >>>>>> cleanup; TEST launch_cluster 2; TEST $CLI_1 peer probe
 >>>>>> $H2;
 >>>>>>
 >>>>>> EXPECT_WITHIN $PROBE_TIMEOUT 1 check_peers
 >>>>>>
 >>>>>> $CLI_1 volume create $V0 $H1:$B1/$V0 $H2:$B2/$V0 EXPECT
 >>>>>> 'Created' volinfo_field $V0 'Status';
 >>>>>>
 >>>>>> $CLI_1 volume start $V0 EXPECT 'Started' volinfo_field $V0
 >>>>>> 'Status';
 >>>>>>
 >>>>>> #Mount FUSE TEST glusterfs -s $H1 --volfile-id=$V0 $M0;
 >>>>>>
 >>>>>> TEST mkdir $M0/dir{1..4}; TEST touch
 >>>>>> $M0/dir{1..4}/files{1..4};
 >>>>>>
 >>>>>> TEST $CLI_1 volume add-brick $V0 $H1:$B1/${V0}1
 >>>>>> $H2:$B2/${V0}1
 >>>>>>
 >>>>>> TEST $CLI_1 volume rebalance $V0  start
 >>>>>>
 >>>>>> EXPECT_WITHIN 60 "completed" CLI_1_rebalance_status_field
 >>>>>> $V0
 >>>>>>
 >>>>>> $CLI_2 volume status $V0 EXPECT 'Started' volinfo_field $V0
 >>>>>> 'Status';
 >>>>>>
 >>>>>> cleanup;
 >>>>>>
 >>>>>> Regards Anand.N
 >>>>>>
 >>>>>>
 >>>>>>
 >>>>>> _______________________________________________
 >>>>>> Gluster-devel mailing list Gluster-devel at gluster.org
 >>>>>> <mailto:Gluster-devel at gluster.org>
 >>>>>> http://www.gluster.org/mailman/listinfo/gluster-devel
 >>>>>>
 >>>>>
 >>>>>
 >>>>> _______________________________________________ Gluster-devel
 >>>>> mailing list Gluster-devel at gluster.org
 >>>>> <mailto:Gluster-devel at gluster.org>
 >>>>> http://www.gluster.org/mailman/listinfo/gluster-devel
 >>>>>
 >>>>
 >>>> If it is a crash of glusterd when you do rebalance start, it is
 >>>> because of FORTIFY_FAIL in libc. Here is the patch that Susant
 >>>> has already sent: http://review.gluster.org/#/c/11090/
 >>>>
 >>>> You can verify that it is the same crash by checking the core
 >>>> in gdb; a SIGABRT would be raised after strncpy.
 >>
 >>
 >> glusterd  is not crashing, but I am getting rebalance status as
 >> fail  in my test case. It is happening in test frame work ( any
 >> simulated cluster environment in same node ) only. RCA: 1. we are
 >> passing always "localhost" as volfile server for rebalance xlator
 >> . 2.Rebalance processes are  overwriting  unix socket and log files
 >> each other (All rebalance processes are creating socket with same
 >> name) .
 >>
 >> I will send patch for this
 > I thought we were already in an agreement for this yesterday. IIRC,
 > the same is true for all other daemons. As of now we dont have any
 > tests which invoke daemons using cluster.rc
 >
      ya .. yesterday we found that volfile server is the  problem , I 
modified volfile server but still i was getting rebalance  status fail . 
Initially I thought some problem in rebalance process, later I found 
that rebalance not able send respond to to glusterd after completing 
rebalance due to unix socket file corruption  and  all rebalance  
daemons are writing log into same log file .
  I think there is no issue with other daemons  which are are using SVC 
framwork work.

patch: http://review.gluster.org/#/c/11210/  - this patch enable the 
writing test cases  for rebalance in cluster environment.


>>
 >> Regards Anand.N
 >>>
 >>>>
 >>> AFAIR Anand tried it in mainline and that fix was already in
 >>> place.  I think this is something different.
 >>>> -- Raghavendra Talur
 >>>>
 >>
 >>
 >


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.gluster.org/pipermail/gluster-devel/attachments/20150613/d67a0253/attachment.html>