Maybe... good approach... ..... These routers have been power-cycled in the past. It should've shown itself then....
I agree that a VLAN was configured from the beginning and that setting did not matter as long as the VLAN ID was the same. I went over every router and found a mix of it enabled and disabled. I'm guessing that the default if enabled with a default vlan id. I suspect that these boxes glitched and the config was partially read back srambled...perhaps the vlan id was gibberish. I would suspect a warm reset to cause this sort of thing, after all, we're dealing with ram and it can be a bit unstable at times. The freaky thing is that the cold boots insured that the config was read from flash. (I know flash is not permanent...maybe its going bad) .. that config was still bad. The log showed no config changes.. What gives... We suspected the nic in the Novell Server was psycho; it was replaced and that didn't fix it... Oh well... Tim Fournet wrote: > My first guess would be that maybe the switches had been configured to > turn VLANs on some time in the past, their configs saved to firmware, > then the VLAN configurations were taken off but not saved. The power > cycles then reverted the switches to that old saved state with the > VLANs back on. Of course, that's just one avenue to check. Don't > forget to rule out something causing a flood of the network, possibly > broadcasts from a buggy box. Just don't make the mistake that a lot of > people do and assume that it's some sort of attack without first > checking out other possibilities and narrowing down the problem. > > -Tim > > CMB wrote: > >> OK, I have one for you... >> >> Today, around 11:30 or so, my BAY 281XX switches and XLR1200 started >> freaking out. No one could talk on network without dropping massive >> packets. Novell Server was reporting that it had an ip conflict with >> itself. >> >> Ethereal on Knoppix-STD reported no weird packets. EtherApe showed >> healthy network activity. Cheops mapped out a partial network. >> (Knoppix-STD really rocks, I brought it up on one of my dual CPU >> machines and was impressed..) >> >> I shut down the net and took up back up slowly; the end of the net on >> the other side of my backbone (LX) ports stabilized but my >> 28XXX switches cycled cpu utilizations up and down ranging from 100% >> to 50% (normal is 5%). It was nasty. I went through those and >> isolated a connection on an expansion port. This sent utilizations >> back to normal but I couldn't route between switches. >> >> Brought up some more and they put the network into chaos. I had to >> power down three of my front office switches and leave up two that >> ran vital services. >> >> All front office machines were offline. >> >> I went to verify the config on the nasties and noticed that all >> front office 281XX switches had VLan enabled. (Even the ones that >> worked) WTF? Huh? Fixed 'em and network started performing >> wonderfully. I am still scratching my head as to how these got >> turned on and why the setting mattered to the router....but I do have >> a theory as to what caused them. >> I'd like to know your ideas? >> >> Keep in mind, these are about 6 or 7 years old (were configured >> before I got there), our main UPS feeding these is in bypass because >> its fried itself (Thanks Toshiba), We've had a couple of unexpected >> power offs, and the electricians were working on light-ballasts today. >> >> >> >> >> >> >> >> >> _______________________________________________ >> General mailing list >> [email protected] >> http://brlug.net/mailman/listinfo/general_brlug.net > > > > > > _______________________________________________ > General mailing list > [email protected] > http://brlug.net/mailman/listinfo/general_brlug.net >
