Kipmi0 eating up to 99.8% cpu on centos 6.4
We have centos 6.4 and the kipmi0 is showing as 99.8%cpu and 0.0% memory and load average is 1.00. What should we do to rectify on this? Thank you.
|
Hi,
Maybe you can find your answer here: http://www.serveradminblog.com/2011/02/kipmi0-problem/ hope it helps... Sire Maxus |
Dear Sire,
I dont find this in my centos 6.4 /etc/sysconfig/lm_sensors ? |
Hi,
Any help or indication how to resolve this matter? |
Hi,
I've been thinking about this, but if you don't have the package lm_sensors installed then the problem must lie elsewhere. can you post more info from the "top" command and perhaps install the package sysstat (yum install sysstat -y) and try some test on your box. try to use iostat and "vmstat 5 5" and post your results here. Good Luck Sire Maxus |
Hi,
What type of test should I run with the sysstat? I am not so clear on that ? How many iostat and top samples do you want? Thank you. |
Hi Sire,
Below are some of my samples data captured.Please let me know if those are not suffice. Quote:
|
Hello,
Your system looks OK, besides kipmi0 using 99.8% (it's only using 1 core not all CPU) In general your system is using 12.5% or 12.6% (average)and sometimes uses .12% on IOWAIT which is not a bad number either. If you could post the exit of this command: #ps -feaux > /root/process.txt and the upload that file, we could check the process kipmi0 and any other process that could be making kipmi0 to use that much memory. Good Luck Sire Maxus |
1 Attachment(s)
Hi Sire,
What I/O wait could be considered as bad or danger? I have uploaded the required file. Thank you and appreciate your help. |
Hi Sire,
Any updates based on the process list? Thank you. |
Hi,
As far as I know, 20-25% is considered acceptable for IOWAIT, more than that signals an issue with the storage devices. Code:
root 129 0.0 0.0 0 0 ? S Apr17 0:00 \_ [pciehpd] Maybe some piece of hardware attached recently or the service didn't update cleanly when you update your system. One question I have not asked yet is if you have rebooted your system after the update? Sire Maxus |
Hi Sire,
I am sorry kind of new into this area. So which of the previous commands best to be used to monitor the IOWAIT. IOTWait signify that there is delay in the harddisk rite? I am sure there is no additionaly hardware attached to this machine. Possible the service didnt update cleanly. How ensure that there is a clean update cause I just run yum update always thats it. Normally I dont reboot. But there was once I reboot last month but after reboot is ok then slowly it again hike to this values. I am curious how do you linked pciehpd to kipmi0? Actually what is the exact role of kipmi0. |
Hello,
The option "f" for the PS command shows all process with their parent and child process. So there you have the process "pciehpd" which is a parent process for the "kipmi0" process, you can see the relationships between process with the lines drawn to the left of the process name. You can check more infor on kipmi in this links: http://www-01.ibm.com/support/docvie...2575fa0050f604 http://lists.us.dell.com/pipermail/l...ay/031305.html https://supportcenter.checkpoint.com...tionid=sk43262 http://www.linux-archive.org/red-hat...el-thread.html I hope this helps you, Sire Maxus |
Hi sire,
Actually I have visited all the given links via google and non of it working e.g. service ipmi stop also is not working. I am quite lost on how to exactly solve this? Do you think a reboot again will help? Any way to ensure a clean yum update? |
Hi,
I thought maybe a reboot could help, but if you have already done it, and then the process keeps hogging CPU then that is not a solution. Of all the links I've read kipmi0 is related to IPMI, which is a set usually used to monitor hardware or used by some applications to monitor some process. From the links I posted on my previous post, the IBM one says that it does not matter if kipmi is reporting high CPU usage, but it only runs on idle time and is standard behavior for this process. If it bothers you that the process uses so much CPU the you can disable it as said in the following link: http://www.novell.com/support/kb/doc.php?id=7003352 (Novel SUSE) http://unix.stackexchange.com/questi...-on-centos-6-4 (CentOS) Good Luck Sire Maxus |
Hi Sire,
IS quite challenging as I have followed on the Centos OS solution but I dont find this file /etc/sysconfig/lm_sensors too. I am practically lost and have no idea what to do next as the disable also not working either. |
Hi,
How about the "ipmi" service in section "workaround" from "Potential cause #1", does it work? In case the ipmi service is not there, What does the "dmesg" command output? #dmesg > /root/dmesg.txt Good Luck Sire Maxus |
1 Attachment(s)
Hi Sire,
How mean service ipmi stop no even this does not work either too. It gives me ipmi:unrecognized service. I have attached dmesg.txt file too. |
Hi,
Looking at your dmesg file, I found that the server is a Cisco UCS C200 M2 (right?) Then, searching the forums for cisco support i've found this: Problem with sensor that is corrected with a Bios upgrade https://supportforums.cisco.com/message/3530456#3530456 I'll let you know if I find anything else. Good Luck, Sire Maxus |
Hi Sire,
Yes you are right we are using Cicso UCS C200 M2. IF we upgrade the bios or firmware I am not too sure will these effect the warranty of the servers what is your opinion on this? |
Hello,
If your servers are still under warranty you can open a support case with cisco and ask them if the bios update would solve the issue you are having right now. Take advantage of your support contract, that may lead to a hardware change if they detect that there is some faulty hardware. Usually support ask you to update bios, CIMC, firmware, etc. and test if that solves any of the issues you have, if not, then they begin to dig deeper. For every update or upgrade you can ask cisco support the procedure according to best practices. Good Luck Sire Maxus |
Hi Sire,
Definitely I will now focus on that but incase you find any clue from Linux do let me know so I can test them too. Thank you. |
Get rid of kipmi0 (CentOS 6.4)
Hi! For some reason I also wanted to get rid of kipmi0 on my CentOS 6.4 x64 installation. It was built in the kernel. Thus I edited kernel .config and replaced all *_IPMI*=y entries by =m. Rebuild the kernel and reboot. No more kipmi0.
Good luck! |
Dear Pethruha,
How do you rebuild the kernel ? What is the =m stands for ? |
Dear newbie14,
I summed up what I did on http://myelectrons.com/build-linux-kernel-centos/ I'd be glad if you let me know whether it was helpful or otherwise. Cheers, - Serge. |
All times are GMT -5. The time now is 06:06 AM. |