[Gluster-users] Help interpreting profile results

[Paul E Stallworth]

Hello everyone,

I have been tasked with helping to find out why we are having issues with our website page load times.  Our webstack consists of 3 apache servers, 3 glusterfs servers, and 3 mysql servers, backed by Nimble storage.  On the glusterfs machines the gluster disk is mounted as ext4 and is mounted as nfs on the apache servers.  None of the configuration decisions are my own, I have inherited this problem from a recently resigned colleague.  I read through some other messages on this list that advised starting the profiler, let it run for a minute, run 'info', let it run for another minute, and run 'info' ago.  I have done that couldn't find anything in the gluster docs about interpreting the results.  So I put the results in a pastebin to avoid dumping 39k worth of logs on the list.  The profile results can be found here http://pastebin.com/DZK0vNQC.  I would greatly appreciate any feedback on these results.  I am for now trying to isolate if the issue is related to our storage system which was upgraded recently or if it is higher up in our architecture.  My team handles OS and up, another team handles VM and below (including storage and network), so I'd like to at least know which direction to head in.  Any other tips or resources that you could send my way would also be appreciated.  

Thanks,
Paul

Paul Stallworth
Housing IT
University of Colorado Boulder
Boulder, Colorado 80309
T: 303.735.6623