Silly question... how many simultaneous file transfers can a 64M memcache handle? And what parameters would affect that?
Scenario: AFS client is an FTP server using memcache. The system routinely has 100 users or more. We're starting to see FTP clients kick back messages like "Stale NFS file handle"... which of course is a lie... but I'm thinking that there is a resource that is being overrun in the cache manager. Currently -memcache -blocks 65536 -chunksize 16 -stat 2800 -daemons 3 -volumes 128 I can't increase the size of the memcache because attempting to valloc more memory results in an oops most of the time. I'm considering adjusting the other parameters like -memcache -blocks 65536 -chunksize 18 -stat 4000 -daemons 5 -volumes 512 Am I going to have problems if I increase chunksize because there will be fewer dcache entries? Are there limits to how high I can set the other parameters when using memcache? I know it's more of an art than a science (at least it seems that way... or maybe voodoo.... ), but if anyone has experience using memcache on a server with a high load, I'd love to hear about it. Thanks! Mike Polek Pictage, Inc. _______________________________________________ OpenAFS-info mailing list OpenAFS-info@openafs.org https://lists.openafs.org/mailman/listinfo/openafs-info
