Hi, 1. Did the process die for a good reason meaning that you maybe limited memory via cgroups or would you consider this a bug? If so it would be very interesting to know more details.
2. Do you have any logoutput of the restart? What should happen: DBServer restarts and connects to the agency. It tries to lookup its internal ID via the user supplied --cluster.my-local-info and should reintegrate iteself into a cluster. This is integral part of our resilience tests which we execute very regulary. You seem to have hit an edge case or seem to have a problem with your setup. In any case logs and startup options would be HIGHLY appreciated to resolve this issue :) Kind regards, Andreas Streichardt Am Dienstag, 6. September 2016 08:13:58 UTC+2 schrieb [email protected]: > > I've setup an ArangoDB cluster on three machines, following the > instructions present here : > https://docs.arangodb.com/3.0/Manual/Deployment/Distributed.html > > Then i injested documents using ArangoImp and it worked perfectly, with > good performance. > > After about a day, One of the Primary DB servers went down with the below > error : > > 2016-09-06T02:44:24Z [29600] ERROR {threads} could not start thread > 'DispatcherStd': Cannot allocate memory > 2016-09-06T02:44:24Z [29600] FATAL cannot start dispatcher thread > > When i restarted the processes present in the DB server machine, it was > not detected by the co-ordinator machine. > > Are there any guidelines to bring up the server, when one of the machine > goes down. > > However, if i remove all the agency, primary and co-ordinator directories, > and re-start all the process from start then arangoDB is working > properly. > -- You received this message because you are subscribed to the Google Groups "ArangoDB" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
