Memory leak on Catalyst 2960 & 2960S - similar to Bug CSCts52797

[m.litzel]

Hello community,

after upgrading about 35 Catalyst 2960 and Catalyst 2960S to IOS 15.0(2)SE2, we experience a memory leak on several switches. After some days / weeks the switches are not accessible via Console/Telnet/SSH/Web any more. Only SNMP seems to work properly.

Attached users do not experience any decrease in service.

Trying to connect to the console, we get following error message:

"% Low on memory; try again later"

The only (temporary) solution is to reboot the switch. The behavior is similar to Bug CSCts52797.

With regards to the Bug notes this bug should only affect Catalyst 2960 with 64MB of RAM and should already be solved with IOS 15.0(2)SE2.

We experience the erroneous behavior with

-WS-C2960-48TC-S      running IOS 15.0(2)SE2

-WS-C2960S-48LPS-L  running IOS 15.0(2)SE2

See also discussion:

https://supportforums.cisco.com/thread/2103327

Could anyone provide an advice to which stable IOS release we could downgrade or if this bug will definetly be fixed in any specific future release?

Thanks

Michael

[stevecox]

I had this happen this morning to 4 2960S switches running version 12.2(55)SE5.

[glen.grant]

  Not sure why cisco has such an issue resolving this issue. This issue has been around since way back around 12.2.35  train.  You would think they would have fixed this by now.

[Chris Fink]

I had this problem too. Turning off macro auto monitor seemed to help (no macro auto monitor). Nevertheless, I have downgraded all of my 2960 plain switches to 12.2(55)SE8.

[jensbofinger]

Had the same issue using version 15.2(1)E1 with 4-stack 2960S in my lab:
ssh/telnet/console not available "% Low on memory; try again later"

Only solution was to restart the stack but problem re-appeared after a few days...

After doing this
- no macro auto monitor
- no cdp running
- snmp completely disabled

stack-uptime is now 7 weeks and a few days.

Only 5 VLANs, 1 Trunk used and still low on memory but at least able to login to the system.

lab_stack01#sh processes memory sorted
Processor Pool Total:   60627156 Used:   56541992 Free:    4085164
           I/O Pool Total:   14680064 Used:   12391292 Free:    2288772
  Driver te Pool Total:     1048576 Used:               40 Free:    1048536

Switch  Ports   Model                              SW Version        SW Image
---------  -------   --------                              ----------------        --------------
*    1      52       WS-C2960S-48FPS-L    15.2(1)E1            C2960S-UNIVERSALK9-M
     2      52       WS-C2960S-48FPS-L    15.2(1)E1            C2960S-UNIVERSALK9-M
     3      52       WS-C2960S-48FPS-L    15.2(1)E1            C2960S-UNIVERSALK9-M
     4      52       WS-C2960S-48FPS-L    15.2(1)E1            C2960S-UNIVERSALK9-M

Anyone already experience with IOS 15.2(1)E2? Wonder if Cisco managed to fix this bug in the meantime...but not really believe in it

[Leo Laohoo]

Avoid the ENTIRE 15.2(1)E IOS.  I stopped testing 15.2(1)E2 after I loaded it on my test switch and ran into this "%%Low memory" error.  

For 2960S, you can use 15.0(2)SE4 or 12.2(55)SE8.  I haven't begin testing 12.2(55)SE9 yet.

[daniele matta]

Hello guys,

I have 2 cluster

one with 4:

WS-C2960S-48FPS-L

one with

WS-C2960S-48FPS-L

WS-C2960S-24PS-L

both cluster are using the software version 15.0(2)SE2 and I have the same problem.

I have checked the bugtool and it confirm that with the 15.0(2)SE2 the bug should have been solved but i can confirm that this software version doensn't solve the issue.

I wonder if anyone of us can suggest a stable update version which i can use without any worries.

Thanks

[Leo Laohoo]

Use 15.0(2)SE4 instead.

[daniele matta]

did you perform any reliable test?

[Leo Laohoo]

Yes.  All my 2960S are running 15.0(2)SE4 with Dot1X and PoE.  Those two alone should generate very high CPU.  

[quang1600]

I just bought (25) Cisco Catalyst 2960S-48FPS-L Switch and it currently has 12.2.55.SE7.  After reading this thread, should I stay with current release?  I noticed on Cisco download section has 15.0.2(SE5) as suggested release.

I also have 3 stacks with (2) WS-C2960S-24PD-L switches that running 15.0(2)SE.  So far, I don't experience this issue.

[Leo Laohoo]

I noticed on Cisco download section has 15.0.2(SE5) as suggested release.

Don't even attempt to use 15.0(2)SE5.  If you need something in the 15.X range, then use 15.0(2)SE4 instead.  

I just bought (25) Cisco Catalyst 2960S-48FPS-L Switch and it currently has 12.2.55.SE7. 

12.2(55)SE7 is stable but there are more improvements (particularly with security vulnerabilities) with the latest, 12.2(55)SE9.  It is your choice to stay with the current IOS version or upgrade to newer ones.  

[Pavel Pokorny]

Hi,

Is there any special recommendation to avoid 15.0.2 SE5 ?

I have been through memory problems with 15.0.2 SE2, and I would like to put in my boxes something more stable than this.

In 2960G there is SE5 recommended by Cisco. Do you know any problems, about this release?

Thank you

[Leo Laohoo]

As I've said 20 days before, 15.0(2)SE4 is a good version for a 2960/G/S.

[Pavel Pokorny]

You also said to "Don't even attempt to use 15.0(2)SE5". I'm only asking why? What is so bad about this release?

Thank you

[Leo Laohoo]

The old "ghosts" that haunt 15.0(1)SE and 15.0(2)SE train were resurrected in 15.0(2)SE5.  CPU and/or memory spikes came back in 15.0(2)SE5.  The only 15.X that has been, so far, successful and stable is 15.0(2)SE4.

And don't even try to load 15.2(1)E.  This particular train will NO LONGER BE CORRECTED.  By the end of June 2014 Cisco will release 15.2(2)E and this will train will be re-written from scratch and (according to what I was told) better quality/control.