Random TCP-timeouts on 8845 phones

[brettlarkins]

Hello,

I need help trying to figure out why phone registrations are randomly dropping and why we are seeing TCP-timeouts in the logs.

All the phones operating at the same location as the CUCM are fine... all the phones operating at remote locations are experiencing sporadic and random tcp-timeouts.

WAN connection between the sites is MPLS, and there is no firewall between the phones and the CUCM.  We are not seeing any issues with the network, and are not having any issues with any other applications on the network (including SCADA devices).

TCP timeouts are sporadic and random, meaning all the phones at the same remote location do not experience the tcp-timeout at the same time, or even close to the same time.  For example, phone#1 may have a timeout at 8am, and phone#2 may have a timeout at 5pm.  And... the phones may go several days between tcp-timeouts, or they may have 2 in the same day.

We cannot seem to find a rhyme or reason as to why & when, and we cannot replicate the issue on-demand.

We also tried flooding phones with several concurrent calls and we were unable to duplicate the issue.

phones are 8845's and 7841's running:

Active Load ID:    sip8845_65.11-7-1-17
Inactive Load ID:  sip8845_65.11-5-1-18
and
Active Load ID:    sip78xx.11-5-1-18
Inactive Load ID:  sip78xx.11-0-1-11

Cisco Unified CM Administration
System version: 11.5.1.12900-21

Any ideas?

[Slavik Bialik]

Same issues happend to me after I upgraded CUCM to 11.5. All 8845 were reseting from time to time. I tried to investigate the logs, saw few strange things like crash dumps that were created from time to time (I saw them when I went into phone's web page) and few other unexplained errors.

Anyway, found a few Cisco bugs that I thought that may be related (but not in 100%) and saw that they were resolved in Firmware 12-0-1-11, so I upgraded all of the phones to this firmware and the issue is gone now, according to my client. I advice you also to upgrade to this firmware.

[brettlarkins]

That's very interesting... How long have your clients been running 12-0-1-11?

[Slavik Bialik]

Something like 2 weeks already. And no complains at all.
Before that, they were complaining every day, because they got complains from senior managers all the time, because they got disconnected from calls all the time. And in the logs of the phones I saw that the phone couldn't register in time, so after the registration time period (150-180 seconds) has past the phone was loosing registration. Couldn't understand why, because those logs aren't so simple to read... But as mentioned, found few cisco bugs reported that may have lead to this issue:
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCve01459
https://bst.cloudapps.cisco.com/bugsearch/bug/CSCvd37384
Found few more, but now I cannot find them again.

[brettlarkins]

Thank you for the info... I am going to suggest to my team that we try this and see if it resolves our issue as well.

[Jordan1212]

I have this issue as well. 

CUCM 11.0.1.21900-11

Cisco 8851 and 8841s on the local fiber loop do not have this issue.  8851s and 8841s across the WAN on GRE tunnels are dropping and resetting randomly.  Placing customers on park or hold drops calls as well randomly.  Seeing LastTimeCMresetTCP and LastTimeRegistrationTimeOut.

We are not seeing issues at sites that still have 79xx phones. 

88xx's are running 11-7-1-17 and 11-5-1ES-5.

We are seeing no packet loss to these branches.  QoS is configured. Not sure where to go.

Happens on 3 different service provider networks on any of our tunnels to the branch.  Replaced routers, switches and phones. Replaced cabling.

Extensive rounds of troubleshooting does not reveal any issue.

[richard.king@live.com]

We are having the same issue with our VPN clients using 8841 desk phones.

Active load is: sip88xx.12-8-1-0001-455.

any advise as to how we would go about troubleshooting this issue since it is random?