Hi Pham,
We had applied the patch on 5.0 GA and observed that the issue is still
observed.
Below are the steps and the apis used in the application to reproduce the issue.
Application :
-> Invoke saCkptInitialize
-> Invoke saCkptCheckpointOpen with create flag and
SA_CKPT_WR_ACTIVE_REPLICA_WEAK.
-> Invoke saCkptCheckpointOpen with WRITE flag
-> Wait for user to press enter ( to invoke failover )
-> Invoke saCkptCheckpointUnlink
-> Invoke saCkptFinalize
Steps to reproduce the issue :
-> Initially start a single controller and payload.
-> Start the other controller, which shall join as standby.
-> Once the standby controller is joining, invoke the application on the
payload. This is such a way that the CKPT apis shall be invoked when CKPT cold
sync is in progress.
-> After a sleep of 20 seconds, induce middle failover and later unblock the
application after which unlink and finalize apis shall be invoked.
The unlink api returns TIME_OUT and the IMM objects are not deleted from the DB
immfind | grep -i Demo
safCkpt=DemoCkpt,safApp=safCkptService
safReplica=safNode=PL-3\,safCluster=myClmCluster,safCkpt=DemoCkpt,safApp=safCkptService
safReplica=safNode=SC-1\,safCluster=myClmCluster,safCkpt=DemoCkpt,safApp=safCkptService
safReplica=safNode=SC-2\,safCluster=myClmCluster,safCkpt=DemoCkpt,safApp=safCkptService
-> If this application is invoked next time, checkpoint open shall return
SA_AIS_ERR_LIBRARY.
-> At this stage, if the application is invoked twice, ckptd segfaults and the
ticket #2011 is raised regarding that.
This issue (#1765) seems to be similar as #247, which has been closed as
non-reproducible. Some times, checkpoint open also gets SA_AIS_ERR_RESOURCES
as mentioned in #247.
-- Srikanth
Attachments:
-
[1765.tgz](https://sourceforge.net/p/opensaf/tickets/_discuss/thread/8ea9d424/d730/attachment/1765.tgz)
(111.5 kB; application/x-compressed-tar)
---
** [tickets:#1765] saCkptCheckpointOpen api call failed and returing
SA_AIS_ERR_LIBRARY after couple of failover**
**Status:** accepted
**Milestone:** 4.7.2
**Created:** Fri Apr 15, 2016 06:26 AM UTC by Ritu Raj
**Last Updated:** Wed May 04, 2016 06:56 PM UTC
**Owner:** Pham Hoang Nhat
**Attachments:**
-
[ckpt_trace.tar.bz2](https://sourceforge.net/p/opensaf/tickets/1765/attachment/ckpt_trace.tar.bz2)
(3.2 MB; application/x-bzip)
setup:
Changeset- 7436
Version - opensaf 5.0 FC
4 nodes configured with single PBE and a load of 30K objects
* Issue observed :
saCkptCheckpointOpen api call failed and returing SA_AIS_ERR_LIBRARY after
couple of failover
* Steps to reproduce:
> Ran couple of failover and observed saCkptCheckpointOpen failed.
> below is the snippet of agent trace:
Apr 15 8:08:50.275115 cpa [28883:cpa_mds.c:0776] << cpa_mds_msg_sync_send:
retval = 1
Apr 15 8:08:50.275128 cpa [28883:cpa_api.c:1043] T4 Cpa CkptOpen failed with
return value:2,ckptHandle:63
Apr 15 8:08:50.275141 cpa [28883:cpa_api.c:1146] << **saCkptCheckpointOpen:
API return code = 2**
> Traces of both controllers and agent trace of payload is attached.
---
Sent from sourceforge.net because opensaf-tickets@lists.sourceforge.net is
subscribed to https://sourceforge.net/p/opensaf/tickets/
To unsubscribe from further messages, a project admin can change settings at
https://sourceforge.net/p/opensaf/admin/tickets/options. Or, if this is a
mailing list, you can unsubscribe from the mailing list.
------------------------------------------------------------------------------
_______________________________________________
Opensaf-tickets mailing list
Opensaf-tickets@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opensaf-tickets
