I have 6 identical IBM blades and all are running RHEL 3, kernel 2.4.21-40.ELsmp. They are dual 2.8 Xeons with 4 GB RAM. Most of the blades that are not in production have a resting load average of 0.00 to 0.08. One particular blade has a resting load average of 0.2 to 1.3. Even when I shut down just about every service it hovers around 0.3. The load looks to be coming from softirq. How can I track it further?
irq looks fine, as does iowait and context switching, at least as far as I can tell.
Here is a snip from top:
16:30:49 up 1 day, 9 min, 2 users, load average: 0.51, 0.65, 0.83 46 processes: 45 sleeping, 1 running, 0 zombie, 0 stopped CPU states: cpu user nice system irq softirq iowait idle total 0.0% 0.0% 0.0% 0.0% 3.2% 0.3% 96.3% cpu00 0.2% 0.0% 0.0% 0.0% 0.0% 0.0% 99.7% cpu01 0.0% 0.0% 0.0% 0.0% 0.0% 0.0% 100.0% cpu02 0.0% 0.0% 0.0% 0.0% 0.0% 0.6% 99.4% cpu03 0.0% 0.0% 0.0% 0.0% 13.0% 0.8% 86.2% Mem: 4093432k av, 763352k used, 3330080k free, 0k shrd, 213936k buff 278660k active, 220840k inactive Swap: 2048276k av, 0k used, 2048276k free 347588k cached
Thanks,
Brad Crotchett brad@bradandkim.net
On Fri, 2006-08-04 at 16:31 -0500, brad@bradandkim.net wrote:
irq looks fine, as does iowait and context switching, at least as far as I can tell.
I don't know what's wrong specifically but, when diagnosing these kinds of issues, I have almost always found that the sysstat package's tools are invaluable.
On Sun, 2006-08-06 at 10:55 -0500, Jason D. Clinton wrote:
On Fri, 2006-08-04 at 16:31 -0500, brad@bradandkim.net wrote:
irq looks fine, as does iowait and context switching, at least as far as I can tell.
I don't know what's wrong specifically but, when diagnosing these kinds of issues, I have almost always found that the sysstat package's tools are invaluable.
Yes, the utilities are very valuable. I have been through iostat, vmstat, mpstat, and sar. I am still not able to pinpoint the cause.
Thanks,