Has anyone else had problems where the M6220 blade switch management interface stops working, typically after a couple of weeks of uptime?
We've seen this freeze on firmware versions ranging from very old 3.x.x.x up to 4.1.1.9. The latest 4.2.0.4 hasn't been out long enough to be prone to the issue, which seems to be triggered by a combination of internal flash access and a uptime of a couple of weeks or more.
We have a lot of M6220 switches, running fw version 4.1.1.9 (currently the second newest available). Every now and then while working on the CLI, the interface freezes so that it stops accepting input, and no more output is printed. At this point, all management methods (except SNMP) become unavailable: Out-of-band CLI (via the CMC module), direct serial connection, and web.
SNMP continues working just fine. Both snmpwalk, snmpget and snmpset continue working.
What seems to trigger the freeze, is operations that access the internal flash storage on the switch. Things like "show running-config" or "copy running-config startup-config" seems to be the most common (or they might even be the sole) trigger for the freezes. Also, triggering the freeze like this seems to become more likely as the switch uptime increases, typically beyond a couple of weeks.
The main concern here is that there's no way for us (nor Dell, presumably) to troubleshoot a switch once it's in the frozen state. I would think the hw/fw manufacturer (Broadcom) has some tricks up its sleeve, but they might not be practical. The sheer number of switches we have could also mean that there's simply a much higher probability of us observing the issue, rather than someone with two or ten of them, so incidents are under reported.
It would help a lot to hear reports from other users who have run into the same issue, which would help us gain some momentum to get it tracked down and fixed.
