5420 unable to update to 33.6.1.5

[Ansley_Barnes]

I'm waiting on my vendor to respond to this (I opened a case with GTAC but apparently we bought these units through a partner that does the support instead), but I was wondering if anyone else had seen this and maybe had a workaround. I've got a pair of stacks - each is one 5420F-16MW-32P4XE stacked with one 5320-48P-8XE. I attempted to upgrade the stacks to 33.5.1.6, but the 5420s throw the error "Error: Failed to install image - mount: can't find /alt//exos in /proc/mounts" and their inactive firmware partition becomes inaccessible - even running "show version images" throws the error "Internal error --> history file open failed!" 

I tried installing the previous firmware (32.7.3.15) onto the inactive partition, but that fails too. It seems like it might have killed the inactive partition's filesystem or something. This happened across both stacks, so it seems like it might be a bug - anyone else seen this?

These stacks are unfortunately remote, so I don't have the ability to easily get to them and run recovery via TFTP/serial. 

[FredrikB-NN]

Yes, it seems the inactive partition has gone bonkers. Is there any particular reason for not staying on 32.7.3.15-Patch1-19? Not that I've seen to many issues but 32.7 is still the safe choice.

To correct this, I suspect you need to do one or more of these:

• reboot
• format the inactive partition (debug via TAC)
• install via USB or TFTP

None of those is risk free, especially if remote.

[Ansley_Barnes]

Thanks for the response. There are a couple of specific reasons for moving off 32.7 onto 33.5.1.6 specifically, largely a couple of bugfixes I've been working with support to resolve over the last year. These stacks are the only ones with difficulty applying the update, out of many stacks globally - but these in particular are the only ones with 5420s in the mix. The issue only seemed to hit the 5420s, and all the 5420s. The 5320s in the stacks took the upgrade just fine.

I've been managing remote switches like these for many years now, but this is the first time I've seen this error.

Apparently for some reason these were purchased through a reseller that does the support instead of GTAC, so when I opened a ticket with GTAC it was rejected. I'm not looking forward to working with the reseller on this only because there are a couple of layers of language translation involved, which always gets tricky when dealing with very niche technical problems like this one.

Depending on the patch contents, I'd probably be able to get away with the next patch for 32.7 on these if I could get the inactive partition formatted/resurrected.