Post #70,082
12/20/02 9:35:26 AM
|
System Load
I'm having a situation where I've got a high system load but no activity. This then causes my batch/at programs not to be released.
Linux: 2.4.18-10smp.
How do I determine what programs are contributing to the system load if they aren't running yet should be?
|
Post #70,092
12/20/02 10:27:40 AM
|
Runnable processes
System load is a measure of processes in the run queue. These are 'runnable' proceses, which may not be running, generally because of blocked IO. \r\n\r\n In 'top' or 'ps aux', these will appear as processes with an 'R' state flag. \r\n\r\n If your system tends toward a high load average, you can modify your atd daemon's load average restriction with the '-l' flag. Modify this in the appropriate /etc/init.d/ (or /etc/rc.d/init.d) script, and restart the daemon. The default is 1.5, which may be too low on some system. \r\n\r\n You can also check for resource starvation -- most likely open files (lsof, fuser), possibly memory, user process table entries, and other resources. \r\n\r\n
--\r\n Karsten M. Self [link|mailto:kmself@ix.netcom.com|kmself@ix.netcom.com]\r\n [link|http://kmself.home.netcom.com/|http://kmself.home.netcom.com/]\r\n What part of "gestalt" don't you understand?\r\n [link|http://twiki.iwethey.org/twiki/bin/view/Main/|TWikIWETHEY] -- an experiment in collective intelligence. Stupidity. Whatever.\r\n \r\n Keep software free. Oppose the CBDTPA. Kill S.2048 dead.\r\n[link|http://www.eff.org/alerts/20020322_eff_cbdtpa_alert.html|http://www.eff.org/alerts/20020322_eff_cbdtpa_alert.html]\r\n
|
Post #70,095
12/20/02 10:44:25 AM
|
Got it
There seems to be a problem with the GBit link to the Sun NFS box causing certain processes to hang in a 'D' state.
ps -elf shows them.
They are unkillable since they are waiting for hardware. They contribute to the load average, which I think is a bug in the calculation logic. They certainly are not runnable.
Required a box reboot. We are swapping the Gbit card.
Thanks.
|
Post #70,101
12/20/02 11:23:48 AM
|
NFS hangs
NFS is notorious for this. \r\n\r\n Not sure if I'm confusing NFS with Samba, but IIRC there's a mount flag that will allow remote file accesses to time out. This should avoid your hung I/O processes in future. \r\n
--\r\n Karsten M. Self [link|mailto:kmself@ix.netcom.com|kmself@ix.netcom.com]\r\n [link|http://kmself.home.netcom.com/|http://kmself.home.netcom.com/]\r\n What part of "gestalt" don't you understand?\r\n [link|http://twiki.iwethey.org/twiki/bin/view/Main/|TWikIWETHEY] -- an experiment in collective intelligence. Stupidity. Whatever.\r\n \r\n Keep software free. Oppose the CBDTPA. Kill S.2048 dead.\r\n[link|http://www.eff.org/alerts/20020322_eff_cbdtpa_alert.html|http://www.eff.org/alerts/20020322_eff_cbdtpa_alert.html]\r\n
|
Post #70,138
12/20/02 2:36:24 PM
|
Re: NFS hangs
Karsten wrote:
Not sure if I'm confusing NFS with Samba, but IIRC there's a mount flag that will allow remote file accesses to time out. This should avoid your hung I/O processes in future.
Soft-mounting NFS risks file corruption. Hard-mounting NFS risks hangs.
Thus: [link|http://linuxmafia.com/~rick/lecture-notes/nisnfs|Nightmare File System]. Voila!
Rick Moen rick@linuxmafia.com
If you lived here, you'd be $HOME already.
|
Post #70,166
12/20/02 5:13:21 PM
|
absolutely excellent link supplied
of the security issue I have run NFS over an ssh tunnel for grins but nfs should only be used in a pristine trusted environment. thanx, bill
will work for cash and other incentives [link|http://home.tampabay.rr.com/boxley/resume/Resume.html|skill set]
You think that you can trust the government to look after your rights? ask an Indian
|
Post #70,225
12/21/02 10:05:15 AM
|
Re: absolutely excellent link supplied
Well of course not! The same applies to splattering X sessions all over the place. NFS is fine in its realm. At least, I've never had any issues.
-drl
|
Post #70,165
12/20/02 5:08:35 PM
|
you are using automount correct?
hat does solve a lot of NFS stale file handle issues also. thanx, bill bill
will work for cash and other incentives [link|http://home.tampabay.rr.com/boxley/resume/Resume.html|skill set]
You think that you can trust the government to look after your rights? ask an Indian
|
Post #70,314
12/21/02 10:56:17 PM
|
No
I believe everyone is barking at the wrong tree. NFS is merely the carrier of this particluar disease.
The Dell Dual Xeon under Linux has problems with every Gbit card I've used, either internal MB based or PCI Intel or Sysconnect.
It gives checksum errors, hangs the box, kills the associated process, corrupts the data, etc, etc, etc.
We've use the variety of Intel drivers, TIGON drivers, Broadcom drivers, etc.
RH just released a kernel update that solved the EXT3 wierd mount last second non-flush write bug, and snuck in a TIGON driver fix that says will stop it from hanging.
I've started working with a dual Xeon IBM box. If this doesn't have the same problems, I'm killing the Dells and replacing them with IBMs
|
Post #70,328
12/21/02 11:36:51 PM
|
understood BUT!!!
if using automount, you cd to nfs mounted directory it gets mounted, you cd out it is unmounted, this removes any issue of keeping a "not in use at this time network connection open issue." This is a good thing unless you are alwas working in a nfs mounted directory IMHO. thanx, bill
will work for cash and other incentives [link|http://home.tampabay.rr.com/boxley/resume/Resume.html|skill set]
You think that you can trust the government to look after your rights? ask an Indian
|