To put it mildly, my experience with fiber channel switches is limited. In actuality my only real experience is from a couple dozen google searches in trying to find the answer to these questions.
Anyway, here's a bit of background on what's going on before I get to the questions:
My company has a McData 4400 fiber channel switch that connects our ESX servers and our backup server to the SAN. We use What's Up Gold to monitor devices on our network and get notices when something goes down/up. We're currently only monitoring the fiber channel switch by ping over the management port.
The problem is that we've started getting random alerts from What's Up Gold saying that the switch has gone down, which is followed by a corresponding "up" alert about a minute or less later. The alerts typically have been happening in the night or over weekends, so by the time anyone came in and saw the alert it was too late. The alerts were thought to be false, until about a week ago I was at my desk when I got the alert. I immediately tried to ping the device myself and saw that it was legitimately not responding to my pings. About 45 seconds later it came back online like nothing was wrong.
Looking through the event logs in the device shows nothing out of the ordinary and the device didn't reboot unexpectedly. For all intents and purposes things were completely normal, except for the 45 seconds where the device was unreachable by ping. The only common thread is that backups are always running when we get the alert about the device going down.
My question is this:
During times of high load/activity, do fiber channel switches put a priority to traffic over fiber instead of ethernet? Or does anyone have any other idea about what might possibly be causing the problem?
Also, I'd love to monitor more than just ping on the switch--interface utilization/cpu load/...really anything that might give us an insight into what's going on with the device. I've not been able to find a MIB that works (so far contacting the manufacturer has proven worthless in this respect) however. Is anyone monitoring a fiber channel switch with management software or by SNMP? If so, what MIBs are you using to get details about the device?
February 28, 2011 3:38 PM
March 4, 2011 7:57 PM