We are in charge of providing “personalized and relevant content to the user” based on rankings and machine learning. I’ve had the great pleasure of working with the Personalization and Relevance Team during the last 10 months. The corresponding code is here.įinally, I also did a script for Nagios/Icinga, for getting this SOLR info from the webscript.This is a short story of how we managed to overcome the stability and performance issues of our Search and Relevance stack.
This is the probe of concept with the results in our corporate Community instance. I did not test it for SOLR 6 (only for SOLR 4 in Alfresco 201602GA). Also a JSON webscript for consuming data via webscript from external tools such as Nagios/Icinga is available. We could get them in a convenient JSON format from Javascript Console, and finally creating a webscript for the SOLR tracking page. The trick for this new page is that there are two SOLR reports that helps obtaining all of this information (SUMMARY and STATUS reports).
#Solr jvm memory monitor full
A good idea is to monitor maxDoc/numDocs - 1.0 parameter, and if it is about 0.4 (for example), it may be a good idea to do a full reindex and optimization. So they consume RAM resources, and mainly, search throughput is lower (much lower). Deleted documents tie up disk space in the index, cause they are marked as "deleted" but they are still there and they are part of the index. Take into consideration that a large number for deleted documents is not healthy for your Alfresco/SOLR system. numDocs, maxDoc, deletedDocs values for each store: They give you an idea of the number of documents indexed and the fragmentation of your index.In addition, there is a known formula to calculate needed JVM memory for SOLR instances. Besides, it is interesting to know about the JVM Heap memory used by each store, to calculate JVM resources in SOLR nodes properly.
#Solr jvm memory monitor free
It's recommended to have enough space for your indices in disk partition, with at least free space equal to your current indices size. SOLR does merging continuosly and needs enough space for merging task.
This will be a good idea for Enterprise Support Tools too (and in this case may be JMX based). This feature would kill Alfresco Community reboots - I often said that I could have studied a second master during years of Alfresco reboots. Hot edit mode for "System Information" parameters.Anyway, I thought and proposed some of them, I do actually here: I had some ideas in mind, but probably too far from my development skills. I knew the project features, and Support Tools saved my day more than once. I use Support Tools a lot in our Alfresco projects, both original Enterprise Support Tools or OOTB Community version one. I was not sure about the projects either, or how I could help. I have been in previous Alfresco meetups, devcons and summits, but this was my first time in a Hackathon. At first I wasn't sure about attending to Hackathon but Francesco Corti and Angel Borroy (MVP) wisely adviced me to go. The event was really amazing (congrats to the organizers and those who made it happen, great event). Some customers told me that last week I was a little bit eventually consistent, and this post is somehow related. I was at the Beecon 2017 in Zaragoza, the Alfresco Developer Conference, time to re-connect, re-charge, and re-commit with Alfresco Community.
0 kommentar(er)
