Once upon a time, Oliver Schulze L. <oliver@xxxxxxxxxxxxx> said:I have 512MB of swap. only 99MB was used this time. The las week, before the system
Hi,
I'm having a strange problem, after 1 week(6 or 7 days) my RH9 server experience
a sudenly very high load.
The load raises to 14, then 20 and even 30.
What is RAM usage like (look at the output of "free")? Something may be
leaking memory, and you start thrashing swap after a week.
got the same symtom, the swap usage was 110MB.
I have 256MB in /dev/hda3 and the rest in 3 swap files in /SWAP, /SWAP2 and /SWAP3
I checked the memory with memtest and the swap partition with 'mkswap -c'
I know. Is really strange. I have RH servers running for 100 days in a row.
I reboot the server and then all works with no problems.
Reboot shouldn't be the answer for anything - I've got Red Hat Linux
servers that have been up for years without a reboot (although that will
soon come to an end when we move our NOC).
I never saw this before.
This server is the master LDAP server, the slave LDAP server has an uptime of 40 days.
But the slave server does not run pop3, so I think openldap has less work.
Every slapd process use like 8MB. Maybe is an I/O problem, but I don't see any load
The only not common thing I'm running is: openldap + nscd
OpenLDAP likes to use a good bit of RAM, and I've heard about nscd doing
odd things, so look at RAM usage. If you are much into swap, use "top"
and sort by resident size (hit capital "M" to sort by RSS) to find the
big RAM user.
in the I/O field in top, and there is no disk problem reported in /var/log/messages
When in high load, the disk works like it was reading/writing from swap.
Maybe I should do a 'swapoff -a' when the high load occurs.
Will keep in touch, next monday will be the way when uptime raises to 7 days.
Thanks to all Oliver
-- Oliver Schulze L. <oliver@xxxxxxxxxxxxx>
