I have 4 cache volumes, 2 gb in size, tuned per Craigs TID. Running
bm3.8sp2a on nw6.5 sp1.1. At least once a week CACHE_3 of the 4 volumes
will fill up and we get errors that it's out of space. If I stop proxy,
delete the contents and then start proxy with -cc I can run a while longer.

There are no hardware errors that I can detect, actually the 4 cache
volumes are spread over 2 LUNS on the SAN (2 volumes per lun).
Tradtional...like I said, tunned per the TID.

Why would one fill up when the others have practically nothing in them?
Shouldn't proxy be balancing the use of these volumes for caching?