What type of queries do you send to Solr, what is the QTime returned per query on an average, and how many queries per second/minute? Try increasing the heap to 512 mb and 1024 mb and test it. Solr 9.6 supports QueryLimits termination which should avoid queries from excessive usage of resources and putting solr into an unresponsive state. Solr 9.5 and older versions have "timeAllowed '' param that can be configured per query and limit long running queries.
On Mon, May 6, 2024 at 5:42 AM Imran Chaudhry <ichaud...@gmail.com> wrote: > Hello Solr users crew, > > I have a single-node Solr 9.x instance running on a hosted virtual Debian > Stable server. > > The server technical specs are as follows: > > 2 vCPU > 4Gb RAM > > The Solr instance has about 180,000 documents which get refreshed once a > day via an automated script. Each day about a few thousand documents may > get added, a few thousand may get deleted. The amount of docs will stay > pretty stable or grow slowly. > > I have a Python/Django app using Solr as a backend for document search. > What I'm finding is that at least once a week, the Solr process gets reaped > by the Linux kernel due to an "out of memory" condition. > > My app produces an Error 500 and > I then have to manually restart Solr. > > My Solr instance is pretty much vanilla, it's configured using the > following params: > > /etc/default# grep -i heap solr.in.sh > # Increase Java Heap as needed to support your indexing / query needs > #SOLR_HEAP="256m" > > /etc/default# grep -i mem solr.in.sh > # Expert: If you want finer control over memory options, specify them > directly > #SOLR_JAVA_MEM="-Xms512m -Xmx512m" > > In the Solr web dashboard the JVM-Memory graph is currently showing: > > JVM-Memory 20.9% / 107.13 MB > Then in grey: > 512.00 MB > 512.00 MB > > What can I do to stabilise things? > > Kind regards, > Imran >
