Solr dead but pid file exists

[brandon-lile]

For some reason, when using these scripts, on the first run, I always get 'solr dead but pid file exists'. It never happens from then on however. I have tested this on multiple boxes (10+), and they all have the same result. Doesn't work initially, but works everytime after.

All are CentOS 6.2 boxes incase you were curious. Any ideas?

[jbusby]

Did you have Solr running on these boxes prior to installing this script? It will look for /var/run/solr.pid when you run the start command, if that file already existed you would get this message.

[brandon-lile]

I'll give it a shot and reply back with my results

[brandon-lile]

There is not a pid file for solr beforehand

[brandon-lile]

This is very strange. It seems as if I have to wait a certain amount of time after running chkconfig. I ran chkconfig through fabric, then ssh'd into the box and manually started it and it worked.

[jbusby]

Can you post the fabric script?

[brandon-lile]

The whole thing is huge. I'll just post the necessary area

# place init script for solr
put('install/solr.initd', '/etc/init.d/solr', use_sudo=True)
put('install/solr.start', '/opt/solr/example/solr.start')
put('install/solr.conf', '/etc/solr.conf', use_sudo=True)
sudo('chmod +x /etc/init.d/solr')
sudo('chmod +x /opt/solr/example/solr.start')
append('/etc/profile', 'export JAVA_HOME=/usr/lib/jvm/java-1.6.0-openjdk-1.6.0.0.x86_64/jre', use_sudo=True)
sudo('sudo chkconfig --add solr')
sudo('service solr start')

The file has already been configured

EDIT*
It's not a matter of time. If I'm ssh'd into the box and run service solr start directly after the fabfile finishes, it works correctly.

[jbusby]

I'm assuming the problem occurs on the last line? I only have a centos 5 VM to test on right now and I can't recreate issue. I have a centos 6 VM at home, I'll have a look tonight. In the mean time if you find the solution please let me know.

[jbusby]

I'm closing this, as I'm unable to recreate it.