0
322
ViewsJob ManifoldCF works, but freezes after some seconds
I have installed ManifoldCF, they connectors and postgres. I have 2 jobs on my ManifoldCF :
- a LocalFile job to external SolR in production
- a JCIFS job to local SolR
On this image, you can see the issue. I can start the job and they index documents but they jobs freeze after many seconds and I can't abort they job if it's freeze. I need to restart manifoldCF to abort and/or restart they jobs if they are freeze. Do you have a solution ?
PS : Time before freeze, is not fix. Number of documents index too. Not correlation. PS2 : I don't have logs for this problem
On logs folder : cat . -name * | grep -R ERROR | grep 2017-04-21
Return just error images tiff on ManifoldCF interface.
1 answers
Answer question0
In cases of deadlocks, we had created separate restful services to either restart agents, restart the Tomcat server on which ManifoldCF was running and also restart zookeeper if we were running that.
Methods to use to avoid such deadlocks: 1. Provide more heap space as well as ram to ManifoldCF. 2. Adjust the number of worker threads in properties.xml and also adjust each job/repo/connector throttle in correlation to your Ram/heap provided. 3. Try to provide more heap to Solr as well, and check its log if that is not causing any issue. Since we faced most of the issues from Solr end rather than ManifoldCF. 4. Try using the multiprocess zookeeper based or multiprocess file based example. 5. Use postgresql / MySQL instead of hsql that is used by default. 6. When using a database, properly configure it according to the instructions given. Also regularly vaccum and verbose.
At the end when the job was stuck , even we had to resort to restarting the agents or Tomcat itself.