I'm trying to take advantage of all the RAM on my glusterfs client
nodes. I'm using the ioc xlator and have set cache-size to 1200MB. At
one point it actually managed to fill 1200mb of cache (I determined
this by looking at memory usage in ps auxw for the glusterfs process).
I decided to bump the cache up to 2048MB, so I made the conf change
and restarted glusterfs. After that, the cache never went past 120mb
(not 1200mb, it never went past 120mb). So then I thought, glusterfs
must not like 2048mb, so I set it back down to 1200mb. Same problem,
still won't go past one-hundred-twenty MB.
The workload is exactly the same, basically reading a lot of large
files. None of the files themselves are bigger than 1GB.
Is there any way to find out what is currently in the cache? Do you
have any suggestions on how to get this working? Below is my conf
file. This is all running on 32-bit CentOS 5 with 2.6.23 kernel,
glusterfs-1.3.8, using fuse.
Any tips will be greatly appreciated.
Dan Parsons
