cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
2925
Views
0
Helpful
3
Replies

CM 10.5 very high CPU usage

Hello,

Please help me understand what logs can I collect ( or trace settings should I do ) to see what generated a very high CPU usage on Call Manager . I can see the moment it went up , but I am not that interested in the CPU monitoring, but rather, understanding the event that caused it.

Thank you

3 Replies 3

Manish Gogna
Cisco Employee
Cisco Employee

Hi,

Here you go

http://www.cisco.com/c/en/us/support/docs/unified-communications/unified-communications-manager-callmanager/118702-technote-cucm-00.html#anc1

Manish

Hello,

Thank you for your help, I didn't find anything in the logs to help me understand what happened, some of the logs got lost because the files were too big, and the logs I've still found reported nothing happened that day. I did see something and I suspect that may somehow lead to this problem, somewhere the client has a few very old phones , 3911 type, that generated a lot of messages registering and unregistering from the server before the CPU load began to raise. Do you think this might be the problem? Has anyone ran into this kind of problem before?

Thank you,

Aseem Anand
Cisco Employee
Cisco Employee

Hi,

You can collect event viewer application and system logs to see what happened. To understand what caused it you need to see event viewer application and system logs prior to the alert, for example you can look for events like low virtual memory, low swap partition. Also, if you have VG224 or any other analog voice gateways, make sure you shut down any unused ports as these analog ports tend to send repeated registration requests in the form of transient connections and can affect CPU performance.

Logs required:

1. Event viewer application logs

2. Event viewer system logs

3. RIS DC perfmon logs

Possible reasons for CPU spike:

1 . Core dumps which indicate a service malfunction.

2. Make sure you are not running DRS backups during the production hours.

3. Check the time of directory sync and see if it matches the time of the high CPU or not.

4. Make sure the vNIC being used is VMXnet3.

5. VMware snapshots are not supported by Cisco. if you have any snapshots i would suggest you to remove them.

6. Make sure the server is synched with an NTP server at a stratum less than equal to 4 and windows based NTP servers are not supported.

Aseem