On 8/27/2014 8:42 PM, hendra_budiawan wrote: > Yes i'm just worried about load average reported by OS, because last week > suddenly server can't accessed so we have to hard reboot. I'm still > investigating what is the problem, because this server is dedicated to solr > only, we suspect the problem came from the solr process but i'm still > looking another possibility what makes this problem arises. Can you give me > suggestion what supposed i need to check further?
What kind of query volume is your Solr server supporting? Are you doing complex queries with a lot of clauses, facets, or something else that's CPU intensive? Is your update volume high? The numbers that you've shown, assuming that the htop info is accurate and you really do have 16 or 32 CPU cores, do not look like any major problem. Solr is working hard, but there's a lot more CPU capacity left. The top output shows that iowait percentage is not a problem, so it's not stuck in disk I/O. Memory usage indicates that OS disk caching is working well. It looks like you were running jetty, but that the jetty might not be the one included in the Solr example. If it's not the one included in the example, then its configuration is not well-tuned for Solr. If you have a high request volume, you may need to increase the maxThreads parameter in the jetty config. The only possible thing that I can think of which might cause a complete inability to access the server via ssh or other means is that you are hitting the open file limit in the operating system. Most linux distros use /etc/security/limits.conf to configure the open file limit for each user. Thanks, Shawn
