[Systems] aslo-proxy update

dfarning at sugarlabs.org dfarning at sugarlabs.org
Wed Nov 25 12:39:35 EST 2009


We had some trouble with aslo-proxy yesterday.  Squid ran out of inodes and crashed.  This brought down activites.sl.o until we figured out what was happening.

Current configuration:
VM 2024MB of memory and 10GB of disk space.

I have tuned squid3.conf more conservatively:
- cache_dir is reduced to 2000MB
- cache_mem is reduced to 1024MB

Long term the solution _maybe_ to rebuild the filesystem with more inodes ( the default is 64K) In the mean time the solution is to keep the cache_dir small enough that old content is removed using a LRU algorithm before we hit the inode limit.

Short term plans:
- monitor inode usage/ cache_dir size
- monitor overall 'memory usage' vs. 'cache_mem' vs.  'memory hits as % of hit requests' and how they affect performance.

Possible infrastructure requests.
- move VM from cow to lvm.
- increase VM memory from 2GB to 3GB
-rebuild filesystem with more available inodes.

david 
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 271 bytes
Desc: OpenPGP digital signature
Url : http://lists.sugarlabs.org/private/systems/attachments/20091125/637f4403/attachment.pgp 


More information about the Systems mailing list