Pete's Log: Silly switch
Entry #2008, (Coding, Hacking, & CS stuff)(posted when I was 43 years old.)
Sigh. I finally have a little time to look at my cluster again and I've found it in a sad state because apparently all my pis rebooted at 5:38 am today and some things didn't come back clean.
Looking in the pi logs, it seems like a hard boot. I checked a few other things on my UPS and their uptime is OK. The PoE switch that powers the pis apparently doesn't want to show me its uptime, but the access point is also powered by the switch and it shows me it also booted at 5:38 am. So looks like my switch had an issue. Going to have to keep an eye on this situation.
I also see in the logs on the switch on my desk that the link to the laundry room switch went down and then up at that time. And that the same thing has happened on September 5 and August 28.
The laundry room switch is made by the same vendor as the desk switch, garage switch, and laundry room access point. But I accidentally managed to buy it as an older hardware revision so its management interface is pretty limited compared to the other switches/APs. Wonder if now I have a second reason to want to upgrade it.
I'm well under the power cap for the switch, so I don't think I'm overloading it. Unless the pis decided to all suddenly draw 3-4 times more power than usual at the same time.
Hmm hmm hmm. So much to ponder. So little time. I guess step one is just try to get my cluster configured so it comes up clean after a hard boot.