On Fri, Mar 30, 2007 at 03:23:36PM -0500, Tony Shadwick wrote: > I think what I'm going to try then today is plug a spare SATA hard drive > into the SR1520, set it up as a raid0, let the CLN pick it up, and > designate it as swap (or I guess I could use LVM and do the same?) and > see if that fixes things as well. If so, then the simplest thing to > instruct CLN users to do is make sure they allot swap prior to > attempting OpenAFS.
I'm not sure that with the Linux kernel as it exists today it's a good idea to swap on AoE (or other network-based) storage, because there isn't yet a good way for the packets that say, "Yes, the data made it to disk," to get processed without using up memory. When flushing out dirty pages to network storage in order to free up memory for other uses, there is a potential for deadlock. Usually you can tell the Linux VM subsystem not to get into a situation where it won't have enough free pages around to receive network packets, but still... It seems like a more likely solution would be to find a good ulimit that works for the OpenAFS process and still catches crazy stack allocations, or else for the OpenAFS code to just switch from stack-based memory to malloced memory. -- Ed L Cashin <[EMAIL PROTECTED]> _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
