Based on the version you're running I suggest you investigate whether
possibly related to SOLR-13336 [1].

Particularly vulnerable configs would have query-time
WordDelimiter[Graph]Filter configured to split and catenate, or
Synonym[Graph]Filter with multi-term synonyms.

[1] https://issues.apache.org/jira/browse/SOLR-13336


On Fri, May 27, 2022 at 9:36 AM Vincenzo D'Amore <[email protected]> wrote:
>
> For my understanding there is no such thing like a feature that kills a
> specific query that would cause an OOM.
> Either in Solr or in general in any other search engine.
> I suggest you start investigating if Solr memory is enough for the kind of
> queries you have, then try to understand what query is dangerous (in terms
> of memory allocation) and why.
>
>
>
> On Fri, May 27, 2022 at 12:15 PM Parag Ninawe <[email protected]>
> wrote:
>
> > Hi,
> >
> > We are currently using Solr 7.7 in Cloud mode
> > The issues we are facing is with Search queries where it takes down
> > the solr processes with JVM OOM issues.
> >
> > A recent incident we faced where a single search query with multiple
> > boosted fields in the query, took down the solr process with
> > out-of-memory issue.
> > I understand there is a proactive way to check such scenarios and
> > never let them happen at the first place but in our case,
> > possibilities are kind of endless at the moment and the Search queries
> > are not completely in our control
> >
> > I was thinking if there would be some feature which can kill the
> > specific query which would cause OOM, so that, we can avoid the Solr
> > processes going down, but seems such inbuilt feature is not present
> > currently in Solr
> >
> > Looking for some help if anyone has built something around it or if
> > someone has done anything to reduce such scenarios
> >
> > I am open to any ideas or suggestions in this matter, Our end goal is
> > to avoid the case where Solr process goes down with OOM due to
> > resource hungry Search queries
> > Any help or direction towards this matter would be immensely helpful
> >
> >
> > Thanks,
> >
> > Parag
> >
>
>
> --
> Vincenzo D'Amore

Reply via email to