Has anyone ever come across UDS causing high CPU conditions on CUCM for Tomcat?
'utils diagnose test' is reporting high CPU for this when my Jabber clients logon in the morning. When I pull the Tomcat logs I can see Errors/Requests/Sessions Active as follows:
Line 396: cucm-uds -> Errors = 15352
Line 397: cucm-uds -> Requests = 1234225
Line 398: cucm-uds -> SessionsActive = 385
I have approximately 900 Jabber devices in softphone mode on the CUCM.
If you are seeing high UDS traffic on CUCM which coincides with the Jabber login may be you would like to check for some points below.
1. Does high CPU happen only on a specific CUCM node?
2. Are all Jabber clients configured to auto login, if yes then you might want to look at the SRV records (_cisco-uds) which direct Jabber to send http requests to CUCM nodes.?
3. What type of connection does Jabber use for Directory? Does Jabber connect to CUCM? If yes you might want to randomize the connection traffic towards CUCM from Jabber clients?
4. You might want to check your CUCM release as there are some defects in CUCM 9.1.X release and Jabber below 11.5.X release for UDS, High CPU and tomcat.
1. It actually was reported on both CUCM nodes (Pub + Sub)
2. Not sure on auto login, will have to check this.
3. Jabber is getting its directory from UDS
4. CUCM version is 184.108.40.20600-26
Did you ever happen to find a resolution to this issue? I noticed this message on one of our CUCM subscribers after running utils diagnose test, and I only ran that because of an RTMT alert I got like this:
Processor load over configured threshold for configured duration of time . Configured high threshold is 90 %
tomcat#1 (89 percent) uses most of the CPU.
For everyone posting on this. ¿What´s your Jabber version?
I have found this bug reported on Jabber instead of CUCM, that triggers something like this.