You are right! Riak was killed by the oom killer on all nodes except the one I was looking at.
Oct 14 18:34:01 gr-node03 kernel: [ pid ] uid tgid total_vm rss cpu oom_adj oom_score_adj name Oct 14 18:34:01 gr-node03 kernel: [31808] 106 31808 5178811 3884730 0 0 0 beam.smp ... Oct 14 18:34:01 gr-node03 kernel: Out of memory: Kill process 31808 (beam.smp) score 955 or sacrifice child (I found that the node I was looking at was made temporarily unreachable by a hosting provider's configuration error and then restarted.) Thanks, Jan ---------- Původní zpráva ---------- Od: Eli Janssen Datum: 16. 10. 2012 Předmět: Re: Riak performance problems when LevelDB database grows beyond 16GB Anything in the system logs or dmesg? With vm.swappiness set to the defaults, the oom-killer could be doing its job a bit too well. On Oct 15, 2012, at 12:10 PM, jan.evangeli...@seznam.cz wrote: > Hi Evan, > > regarding the swappiness and disk scheduling: these were set to default, I > will correct it and run another test. > > The hosting provider sets the computer with software RAID1 over 2 physical > disks, do you think it is useful with Riak? > ... _______________________________________________ riak-users mailing list riak-users@lists.basho.com http://lists.basho.com/mailman/listinfo/riak-users_lists.basho.com
