Posted by tripper54 on February 8, 2011 at 5:50am
I've got an install of mercury 1.1 on an ec2 tiny ubuntu instance. All seems to be going fine, except apache solr. Solr seems to quit periodically. When I run a search on the drupal site, I get a message saying "drupal can't connect to apache solr". If I restart tomcat6, the search runs OK, but then fails again after about half an hour.
I had a look at the tomcat log files, here is the last line form /var/log/tomcat6/catalina.out
Feb 8, 2011 5:20:51 AM org.apache.solr.core.SolrCore execute
INFO: [] webapp=/default path=/select params={mlt.minwl=3&mlt.fl=title,ts_vid_3_names,ts_vid_4_names&mlt.mintf=1&mlt.maxwl=15&mlt.maxqt=20&json.nl=map&wt=json&rows=5&version=1.2&mlt.mindf=1&fl=nid,title,path,url,ss_image_relative&start=0&q=id:gbov7w/node/47&qt=mlt&fq=(nodeaccess_all:0+OR+hash:gbov7w)} status=0 QTime=10Is there anywhere else I should be looking for error messages?
Does anyone have any idea what might be going on?
Thanks!
Comments
I had the same issue, I have
I had the same issue, I had to give up on 1.1, went with a Lamp, APC, Mod_pagespeed, boost combo on 1024mb ram. Works really well. Can't wait for things to get to a release point on this project, but things are in beta, so you can't really expect to not have issues.
Hoow about 1.2 ? Did you get
Hoow about 1.2 ? Did you get any good progress with that? I am having aegir failures and may need to fall back to pantheon, or in the worse case gl the root you have gone. I am useing so much time to get a good stable platform, and it has only lead to dead ends.
www.thecenterofthenet.com
http://cn.linkedin.com/in/roberpteatonjr
http://www.facebook.com/robertp.eaton
http://www.alamy.com/stock-photography/6A42F1A1-E40A-48FD-8C7F-FF258A401...
Memory in kB
I think this has to do with the memory config setting which defaults to kB, you should try setting the memory limit in the /etc/mercury/server_tuneables to 96m (or something else, but notice the 'm'). See http://groups.drupal.org/node/96144.
Memory is set to 96m
Thanks for the suggestion, marcus_w. I checked /etc/default/tomcat6 and found it to be as it should, ie
JAVA_OPTS="-Djava.awt.headless=true -Xmx96m"
My setting in /etc/mercury/server_tuneables is blank , so tomcat is using the defaults.
My situation is a little different to thread you linked to. Tomcat starts OK, and solr works for a while, but seems to stop periodically. I think it's each time cron is run.
Is 96MB sufficient? Could this be some sort of memory error? I don't really know what I'm doing, but the log file entry I posted above just seems to be a call to the executable, not an error message as such, so I'm really guessing as to what might be the problem.
Has anyone else had success with solr on an EC2 tiny instance? Could the lack of server resources be the problem?
Cheers,
Phil
moved to large instance, solr is fine now
Hi,
For the benefit of others, this problem stopped happening when I moved my install to an EC2 "large" instance (7.5GB RAM, 4 EC2 Compute Units). It's running solid as a rock now.
I guess the EC2 tiny instance doesn't have the resources to support solr.