Cisco Bug: CSCul18667

[Rohit Khajuria]

HI All

We have been facing an issue in our cluster where UCCX cluster has failed over from uccx-1 to uccx-2 as the master . and after some troubleshooting we found that Cisco Bug: CSCul18667 is causing the issue.

Now, Cisco Bug: CSCul18667 - CUIC doesn't purge JMX logs, where CUIC doesn't provide any mechanism to control the size for JMX logs and doesn't delete the logs under /opt/cisco/cuic/logs/jmx/ as part of log purge.

So, I just want to know if this Bug can literally create any issue regarding the fail over from uccx-1 to uccx-2 as the master.

Regards

Rohit

[Deepak Rawat]

Rohit,

Yes the engine failover can happen due to the defect CSCul18667. Below defect will shed more light on this:

https://tools.cisco.com/bugsearch/bug/CSCun54915/?reffering_site=dumpcr

Below is what actually happens:

1) Disk/logging (Common Partition) usage on the server goes to full capacity due to the defect CSCul18667

2) This in turn creates an Atomikos issue inside the JVM of CCX Engine where the epoch file gets corrupted causing the defect CSCun54915 to come into picture

3) Now due to the defect CSCun54915, CCX Engine goes in shutdown state causing the failover to happen

Regards

Deepak

- Rate Helpful Posts -

[Rohit Khajuria]

Hi Deepak

Thanks for your comment and help !!

But the bug you mentioned about is CSCun54915.

And I am asking about CSCul18667 where CUIC log purge doesn't clean up logs under JMX folder and I am really not sure if it causes CCX Engine to go in shutdown state causing the failover to happen.

Regards

Rohit

[Deepak Rawat]

Rohit,

It looks like you have not read properly what I wrote or maybe you did not understand it. No issues, one more time I will explain this

Since the CUIC log purge doesn't clean up logs under JMX folder (mentioned in defect CSCul18667), this caused the common partition to go full. It can be verified by running below command on UCCX server during the time when issue happens:

show status

The output of above command will show something like below

Disk/active         27093228K       16177128K       10640852K (40%)
Disk/inactive       26631156K       15787220K       10569016K (41%)
Disk/logging        92746972K        1139180K       86820424K (99%)

Due to the common partition getting full, the epoch file inside the CCX engine gets corrupted and cause the Engine to go in shutdown state (for which we have filed CSCun54915) that actually triggers the failover.

Regards

Deepak

- Rate helpful Posts -

[Rohit Khajuria]

Thanks for the explanation Deepak.

Now there is one more thing if you can help me with. How can we stop/avoid this issue to occur ?

I know one way where we need to manually issue a command to delete the CUIC active log files and that too once in a week or two weeks. But is there any other way to avoid the same automatically ?

Regards

Rohit Khajuria

[Deepak Rawat · Accepted solution]

It was supposed to be fixed in 9.0(2) SU3 earlier but that does not show up in the Known Fixed Release section, it does not show any version for that matter.

Also, this defect had been opened for the same issue affecting version 10 and 11 as well with no known fixed release

https://tools.cisco.com/bugsearch/bug/CSCuv76430

Hence contact TAC to check which version provides the complete fix for this defect and then upgrade accordingly. In the meantime, you can delete the log files using the CLI command that you are using, no other workaround is there

Regards

Deepak

- Rate Helpful Posts -

[Rohit Khajuria]

Hi Deepak

Thanks again for your reply and for your help ofcourse !!!

Regards

Rohit

[Deepak Rawat · Accepted solution]

No issues Rohit.

Please remember to mark the thread as Answered so that it can stay on top and help others with same kind of issue in the Community

Regards

Deepak