Anyone else on host49 noticing incredibly slow disk performance? I'm not swapping, not even close to it. Plenty of IOtokens in /proc/io_status too. But any access of the filesystem (vi, ls, running a perl script) takes up to several seconds to complete. I ran hdparm -t and it was < 2MB/s, not great but not too terrible either. Munin graphs show that all the CPU time I'm using is for IOwait.

I don't think that there's much I can do to avoid this. Can the offending node be found and corrected?

Dreaming of Xenode...
- Brian

Mine is on host49 too. My (false) impression is it became noticeably slow during the past few weeks. My loading is always 0.1 or something. So my guess is someone on the host is thrashing the disk....

ywliu

I think it has been slow for at least a week or two, but it seems to be getting more noticeable now. At least to me and I watch my server stats pretty closely.

Guess there's nothing to do but wait for Xen...

I found one node slightly thrashing. Let me know if that improves things.

-Chris