Brocade restart after x amount of days

[ben horan]

I remember a while ago we discovered that there was a known bug where the Brocade 12000 switches would restart after reaching a certain number of uptime days.

Can anyone tell me what his number was - I want to check out switches to ensure we don't get caught out by this.

[TTr]

It was 49 days. I had the IBM label brocades and the bug got me back in 2005.

http://www.redbooks.ibm.com/abstracts/tips0541.html?Open

[Sivakumar MJ._1]

TTr is right.

Please check the release notes..

[ben horan]

We are on 5.0.5f code at the moment although when we encountered this issue last we were on an earlier release but still in the 5.0.x ballpark.

As i remember it, it was well over 400 days uptime.

I've checked the 5.0.5f release notes but can't find mention of it being fixed or a known bug.

[TTr · Accepted solution]

The URL I sent you above has the 49-day and another one, a 497 day ticking bomb along with their respective brocade bulletins. You probably have to login to the brocade support site to find the bulletins and see which vOS release had the fix in it.

[ben horan]

497 sounds like the ticking time bomb I was thinking of. Looks like we are safe given that we are at 5.0.5f release.

many thanks

[Bill Rothanburg]

The problem was with the v4.x versions of FOS. v5.0.5f is the last release for the Brocade 12000 switches.

[Uwe Zessin]

Looks like another one[*] who did not think about the wrap-around of a counter:

a 10 ms counter:
(2**32)/100./60/60/24 = 497.10

a 1 ms counter:
(2**32)/1000./60/60/24 = 49.71


[*] I've seen such a problem about 15 years ago on a OpenVMS MicroVAX-3400 system that did run for that long (497 days, not 49 ;-) without a reboot and then locked-up hard so we had to do a power-cycle.