solr service stops every day

[Peter_Adam_SB]

OS: CentOS 7

Alfresco: 6.2.0 Community

solr: Alfresco Search Services 1.3.1 using Apache Solr 6.6.5

Java: java-11-openjdk-11.0.10.0.9-1.el7_9.x86_64

Every day the solr service stops working. The only message I was able to get in it's solr.log is "ERROR (qtp947679291-17) [ x:alfresco] o.a.s.h.RequestHandlerBase org.apache.lucene.util.automaton.TooComplexToDeterminizeException: Determinizing automaton with 15226 states and 30003 transitions would result in more than 10000 states."

In solr_oom_killer-8983-2021-05-13_17_38_52.log:

"Running OOM killer script for process 2423 for Solr on port 8983
Killed process 2423"

The issue started last week, indexes has been rebuilt since then, no effect. What can cause this error?

[angelborroy]

Do you have additional log?

It should be required to identify the operation that is provoking this exception.

[Peter_Adam_SB]

I have found this in catalina.out three minutes before:

2021-05-13 17:35:08,540  INFO  [webscripts.connector.RemoteClient] [https-openssl-nio-8443-exec-18] Exceptio
n calling (GET) http://localhost:8080/alfresco/s/slingshot/search?facetFields=%7Bhttp%3A%2F%2Fwww.alfresco.o
rg%2Fmodel%2Fcontent%2F1.0%7Dcreator%2C%7Bhttp%3A%2F%2Fwww.alfresco.org%2Fmodel%2Fcontent%2F1.0%7Dcontent.mi
metype%2C%7Bhttp%3A%2F%2Fwww.alfresco.org%2Fmodel%2Fcontent%2F1.0%7Dcreated%2C%7Bhttp%3A%2F%2Fwww.alfresco.o
rg%2Fmodel%2Fcontent%2F1.0%7Dcontent.size%2C%7Bhttp%3A%2F%2Fwww.alfresco.org%2Fmodel%2Fcontent%2F1.0%7Dmodif
ier%2C%7Bhttp%3A%2F%2Fwww.alfresco.org%2Fmodel%2Fcontent%2F1.0%7Dmodified&filters=&encodedFilters=&term=0096
5806*&tag=&startIndex=0&sort=&site=&rootNode=alfresco%3A%2F%2Fcompany%2Fhome&repo=true&query=&pageSize=25&ma
xResults=0&noCache=1620920108282&spellcheck=true&highlightPrefix=%02&highlightPostfix=%03&highlightFields=cm
%3Aname%2Ccm%3Adescription%2Ccm%3Atitle%2Ccontent%2Cia%3AdescriptionEvent%2Cia%3AwhatEvent%2Clnk%3Atitle&hig
hlightFragmentSize=100&highlightSnippetCount=255&highlightMergeContiguous=false&highlightUsePhraseHighlighte
r=true&alf_ticket=TICKET_29198e4a927c3d4ee4066959751844e01f143a02

[angelborroy]

You could try to increase default value for maxDeterminizedStates (https://solr.apache.org/guide/6_6/tokenizers.html#Tokenizers-SimplifiedRegularExpressionPatternToken...) but not sure why this exception is happening.

[Peter_Adam_SB]

I have only the following tokenizers in my /opt/alfresco-6.2.0/alfresco-search-services/solrhome/alfresco/conf/schema.xml config file:

• solr.KeywordTokenizerFactory
• solr.StandardTokenizerFactory
• solr.ICUTokenizerFactory
• org.apache.solr.analysis.WordDelimiterFilterFactory
• solr.JapaneseTokenizerFactory
• org.apache.lucene.analysis.th.ThaiTokenizerFactory

Should I replace one (for example, for the Hungarian language)? Or look somewhere else?

[Peter_Adam_SB · Accepted solution]

I have changed the SOLR_JAVA_MEM="-Xms1g -Xmx1g" setting to SOLR_JAVA_MEM="-Xms1g -Xmx4g" in the solr.in.sh file.

Today there were no stops.

solr status shows now stable 1.1 GB used out of 2.3 GB.

[EddieMay]

Hi @Peter_Adam_SB 

Great news you got a fix & thanks for updating your thread - super useful to other users.

Best wishes,

[erkut3512]

Where is the file location of solr.in.sh ?  I couldnt find it. Can you help please ?

Alfresco 5.2

[abhinavmishra14]

It depends on what version of solr you are using. For solr6, it is available at the root of the installation path. 

For example if you have installed solr6  (ASS 1.0 to ASS 1.3 is supported for ACS5.2.x) in:

Windows: C:\alfresco-search-services-1.3.0, then solr.in.cmd can be found: C:\alfresco-search-services-1.3.0\solr.in.cmd

Linux: /usr/local/alfresco-search-services-1.3.0 then solr.in.sh can be found: /usr/local/alfresco-search-services-1.3.0/solr.in.sh

If you are using solr4, then the java memory options needs to be set in <TomcatInstallDirectory>\bin\catalina.bat or <TomcatInstallDirectory>\bin\catalina.sh

Example:

catalina.bat:

set JAVA_OPTS=%JAVA_OPTS% -Xms1g -Xmx4g

catalina.sh:

JAVA_OPTS=%JAVA_OPTS% -Xms1g -Xmx4g