Ciao,
i work with GLASS 3.1.0.4 and GemTools Launcher - >=1.0-beta.8.7 Sometimes when i do the Admin / Backup command, from GemTools, after some time the system erase the error: a RepositoryError occurred (error 2734), Request for Backup gclock by session 14 denied, reason: vote state is voting, waiting for promote dead AbstractException >> _signalFromPrimitive: (envId 0) Repository >> _fullBackupTo:MBytes:compress: (envId 0) Repository >> fullBackupTo:MBytes: (envId 0) Repository >> fullBackupTo: (envId 0) JadeServer >> fullBackupTo: (envId 0) GsNMethod class >> _gsReturnToC (envId 0) A this point i need to restart the GLASS environment and after, resubmit the backup. I don't understund because it. Any considerations ? Thanks, Dario _______________________________________________ Glass mailing list [hidden email] http://lists.gemtalksystems.com/mailman/listinfo/glass |
It looks like the backup failed because the GC Lock was not available. When that happens what do you get from the following:
System sessionIdHoldingGcLock. System voteStateString. Repository-wide garbage collection is a complex area. Have you watched any of the architecture videos at https://programminggems.wordpress.com/2010/02/05/scaling-objects-videos/, especially #20 and #22 on garbage collection? Do you have logged-in sessions that are idle and not committing? This would hold up voting. If you stop all the sessions without restarting the stone and try again a few minutes later does it work? James On Jun 5, 2014, at 9:48 AM, Dario Trussardi <[hidden email]> wrote: Ciao, _______________________________________________ Glass mailing list [hidden email] http://lists.gemtalksystems.com/mailman/listinfo/glass |
Dario, In the cases where you get this problem do you know if an MFC has been done in the recent past? Do you have EpochGC enabled? I would like to bullet-proof the backup logic, but I haven't seen this particular error in my usage of the system ...
Dale On Thu, Jun 5, 2014 at 10:29 AM, James Foster <[hidden email]> wrote:
_______________________________________________ Glass mailing list [hidden email] http://lists.gemtalksystems.com/mailman/listinfo/glass |
Hi, Yes this happens if try to do a backup right after (few minutes) a MFC. Gerhard
On Thu, Jun 5, 2014 at 8:34 PM, Dale Henrichs <[hidden email]> wrote:
_______________________________________________ Glass mailing list [hidden email] http://lists.gemtalksystems.com/mailman/listinfo/glass |
Hi again, Well during the MFC i hat a second session (Jade) opened, this session is in a long transaction. After closing the Jade session, backup runs through. Gerhard On Fri, Jun 6, 2014 at 6:39 AM, Gerhard Obermann <[hidden email]> wrote:
_______________________________________________ Glass mailing list [hidden email] http://lists.gemtalksystems.com/mailman/listinfo/glass |
Gerhard, Yes, the vote happens right after an MFC ... if all of the gems are active (aborting/committing at fast pace) then the voting will happen quickly, but if a session or two is idle or a topaz, gemtools, or session is logged in (and "not responsive", then the vote can take longer ...
I should be able to detect this in the backup logic ... before attempting the backup ... and provide a bit better message ... Dale
On Thu, Jun 5, 2014 at 9:49 PM, Gerhard Obermann <[hidden email]> wrote:
_______________________________________________ Glass mailing list [hidden email] http://lists.gemtalksystems.com/mailman/listinfo/glass |
Free forum by Nabble | Edit this page |