6509 switch got rebooted after executing sh tech command.

[rishant123]

Hi,

One of our 6509 switch got rebooted/crashed when we execute the sh tech command on it. Earlier we were getting below mentioned error:-

name_svr.proc[65]: Could not register interest for /registry/2992898759/3457843965 : Not enough memory<31>SLOT0:1y42w:;1288733136.95 :

Can somebody please suggest anything on this ,why the switch got rebooted after executing sh tech

[Leo Laohoo]

He he he ... You got bitten too?

Let me see ... 1 unit of 6500 add "sh tech" command and you get a recipe for a disaster. 

[rishant123]

Yes you are right ..any suggestion to get any other reason why it got rebooted

[Leo Laohoo]

In early IOS', "sh tech" can cause the Sup's to crash.  Newer IOSs have this under control.  So, for obvious reasons, what is your IOS version?

[rishant123]

sup-bootdisk:s72033-ipservices_wan-vz.122-18.SXF11.bin

[Leo Laohoo]

The only thing I can get from this version of IOS is Bug CSCsb80803 but it doesn't seem to cause a crash.  You'll get traceback when you do a "sh tech" when using an SSH session.

And boy!  That's a pretty old IOS.

I'd load a newer IOS into  the Sup cards and if the card crashes again, hopefully it will read the newer IOS.

[gephelps]

Without looking at the crashinfo file we cant be sure, but this is probably due to CSCso85859. This bug only affects modular IOS and is fixed in 12.2(33)SXH3 and later releases.

When running modular IOS the command show memory detailed is included in the output of show tech-support.

[rishant123]

can someone tell me how can we check memory leakage in 6500 switch?

[gephelps]

Most of the time it is a matter of collecting output and tracking the numbers. Lets say the device ran out of memory in 2 weeks. After the device boots again and is in a steady state take a snapshot of the memory outputs. Then after a couple of days take another snapshot. You would then look to see how much the free memory declined for the interval and track down the process which is then holding that memory. Repeat the process again after another time interval. If there is a leak then most of the time a pattern will emerge assuming the trigger is still in place.