On 06/27/2013 09:33 PM, Marc Grimme wrote:
With heavy load I mean that the ldap process is consuming MUCH more CPU usage
then before.
As you can see from the top screenshot is that I/O Wait is 54.4% and CPU around
50% this is way too much. And this is not good.
Before it was always around 0-1%.
top - 21:09:53 up 6 days, 4:18, 2 users, load average: 1.73, 1.71, 1.74
Tasks: 107 total, 1 running, 106 sleeping, 0 stopped, 0 zombie
Cpu0 : 37.5%us, 1.9%sy, 0.0%ni, 0.0%id, 54.4%wa, 0.0%hi, 0.0%si, 6.2%st
Mem: 1922724k total, 1547748k used, 374976k free, 133928k buffers
Swap: 2064376k total, 1812k used, 2062564k free, 233944k cached
PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
32134 dirsrv 20 0 1626m 652m 16m S 35.8 34.8 66:33.38 /usr/sbin/ns-slapd
-D /etc/dirsrv/slapd-CL-ATIX -i /var/run/dirsrv/slapd-CL-ATIX.pid -w /var/run/d
912 root 20 0 314m 47m 7220 S 5.3 2.6 0:02.11 /usr/bin/python
-E /usr/sbin/ipa-replica-manage list
2012 root 20 0 192m 5280 1536 S 0.3 0.3 3:43.13 /usr/sbin/snmpd
-LS0-6d -Lf /dev/null -p /var/run/snmpd.pid
1 root 20 0 21304 1352 1092 S 0.0 0.1 0:06.61 /sbin/init
2 root 20 0 0 0 0 S 0.0 0.0 0:00.00 [kthreadd]
you only have one cpu, and the load it is getting is too much for it.
With one cpu, you should see a load under 1, with 2, under 2, etc. For
an explanation on what you are seeing there, take a look at this:
http://en.wikipedia.org/wiki/Load_%28computing%29#Unix-style_load_calculation
When the cpu is in i/o wait state that means the processor has to
wait until it gets data for processing. It *usually* is storage and/or
network contention.
A few questions to get you troubleshooting off the top of my head:
Is this a vm or a real hardware server? What are the specs? Is your
network being used now more heavily than before the upgrade?
Where is the storage? DAS, NAS, SAN? Is it enterprise grade storage,
slow sata disks, fast sas disks? If network storage (nfs, iscsi): what
speed is your storage network switching layer? Is it isolated from the
normal network switches? Fiber channel? What speed are the switches?
If it is a vm: are the disks properly aligned to the specs of your
storage vendor?
Be aware that load tuning is a very complicated business. If this is a
vm and you can allocate another cpu to it, you could try that. But if
the underlying host hardware is memory/cpu/disk/network constrained, it
will not help much. Lots of variables to take into account, I am afraid
--
groet,
natxo
_______________________________________________
Freeipa-users mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/freeipa-users
