On 6/10/2020 12:13 PM, Ryan W wrote:
People keep suggesting I check the logs for errors. What do those errors
look like? Does anyone have examples of the text of a Solr oom error? Or
the text of any other errors I should be looking for the next time solr
fails? Are there phrases I should grep for in the logs? Should I be
looking in the Solr logs for an OOM error, or in the Apache logs?
Are you running Solr on Windows? If you are, then a Jave OOME will NOT
cause Solr to stop. On pretty much any other operating system, Solr
will terminate when OOME occurs. This termination will create a
separate logfile, one that contains very little actual information,
really the only thing it says is that the oom killer script was
executed. That logfile will have a filename like the following:
solr_oom_killer-8983-2019-08-11_22_57_56.log
If OOME is the reason Solr stops running, then the only place that
exception will be logged is solr.log as far as I know ... but there
exists a very real possibility that it won't actually be logged. It
could occur at a place in the code that does not have any logging.
At the URL below is an example of a logged OOME on a Solr server. In
this case, it wasn't memory that was exhausted, the error was logging an
inability to start a new thread:
https://paste.apache.org/aznyg
Thanks,
Shawn