It is usually recommended to run all but the smallest non mission critical solr installs on their own hardware or at least own VM. Besides issues like this, the lucene library is optimized to use system level disk caching and the more other stuff competing for disk, the slower lucene (and therefore solr, elastic, opensearch) will be. Databases are definately not polite neighbors and likely also impacted by solr. Obviously if datasets and query rates are very small on both, and occasionally having issues like tthis is fine, do whatever you like... lastly, Linux installs are typically more common and other folks are more likely to have found any given bug before you. On windows, you are more likely to be the first to find an os specific issue.
On Tue, Jan 28, 2025, 4:05 AM Yaşar Arabacı <yasar11...@gmail.com> wrote: > I was able to get procmon and wireshark records today. It looks like a > windows service (installed by one of the programs we are using) starts > gbak.exe at the exact time solr is shutting down. Attempts to restart > solr are also failed while that program is running. > > Google search reveals gbak.exe is a backup utility for firebird sql > server (which was also installed by the same program we are using). I > found the earlier backups created by the same program and they match > the earlier shutdown times. > > I don't know how and why that program interacts with solr but it gives > me some peace of mind knowing what is causing the problem. > > Best Regards, > > Alexandre Rafalovitch <arafa...@gmail.com>, 27 Oca 2025 Pzt, 06:51 > tarihinde şunu yazdı: > > > > Network setup? Has a firewall appliance? It may drop the connection if > > there is no traffic for a while. E.g. between the nodes. That may cause > all > > sorts of fun events. > > > > And if you were watching it and maybe doing some test requests, that > resets > > the timeout value. > > > > Something like wireshark may catch this kind of situations. Both actively > > by examining the live connection and seeing the device IDs participating > in > > it (because firewall will be a piggy in a middle) and by checking if > > connection was aborted just as Solr shutdown trying to use it. The > timeout > > itself will be silent, it is attempt to use connection after will get TCP > > Fin (or some such). > > > > JDBC pools used to get this issue a lot. So if all else fails, check this > > option. > > > > Regards, > > Alex > > > > On Mon, Jan 27, 2025, 2:48 a.m. Yaşar Arabacı <yasar11...@gmail.com> > wrote: > > > > > Checking the task scheduler and system event logs didn't reveal > > > anything interesting. I also ran procmon during the time it was > > > supposed to shutdown, but it didn't shut down this time. I think this > > > is one of those quantum bugs that only exists when not observed. > > > > > > I will run procmon again tomorrow and post updates if anything > > > interesting comes up. > > > > > > Best Regards, > > > > > > Dmitri Maziuk <dmitri.maz...@gmail.com>, 25 Oca 2025 Cmt, 22:25 > > > tarihinde şunu yazdı: > > > > > > > > On 1/25/25 12:15, Yaşar Arabacı wrote: > > > > > Hi, > > > > > > > > > > I am facing a weird problem where solr shuts down everyday at > exactly > > > > > the same time for no apparent reason. I don't know how to begin to > > > > > debug this. I am sharing some system info and log entries below. I > > > > > would appreciate any tips and insights into understanding this > issue. > > > > > > > > Did you check the task scheduler? And system event logs? > > > > > > > > Dima > > > > > > > >
