B3 dies without errors

[Cheeseboy]

It just means the receiver has not logged in and looked at his/her messages yet.
No, no problems (at least with receiving messages).

[DanielM]

It just means the receiver has not logged in and looked at his/her messages yet.
No, no problems (at least with receiving messages).

Ah. That's just plain stupid. So the message gets sent the second the receiver reads it? Logic...

/Daniel

[Cheeseboy]

*cough cough*

Daniel, can you enlighten me a bit? Five posts up?

[DanielM]

*cough cough*

Daniel, can you enlighten me a bit? Two posts up?

Well, the reason I answered is I'm not 100% sure, but I've also interpreted it as you, that if there's a softlink in that directory the module is active and if not it isn't. But don't hang me if I'm wrong

/Daniel

[Ubi]

yeah you remove the softlinks in /etc/munin/plugins. Only thos that remain are used. Test yourself by telnetting to localhost port 4949.

[DanielM]

Ok. This is a weird coincidence. My B3 actually crashed this morning. By the look of ubi's page I would say it happened sometimes between 5 and 6. So I guess it's time for us to start analyzing data, right?

Can't really say it seems like the B3 was working hard. I had two ongoing torrents, hence the network traffic. But nobody was awake when it happened, so wlan was practically idling.

/Daniel

[Ubi]

yeah, about 5:10 that it crashed. But I see no weirdness in the system whatsoever. Really odd...

[DanielM]

Just out of interest, ubi: What happened between 14:30 and 13:10 today?

Oh, and btw me and Tor had a little discussion today today about him helping me to setup console port logging of my B3. Seems like the only way to find out what's really happening here...

/Daniel

edit: Been looking more closely at the graphs now, comparing them to ubi's and cheeseboy's. One thing I notice is that my system was much higher on swap before the crash than what seems to be "normal". Any chance this could be relevant?

[Ubi]

Just out of interest, ubi: What happened between 14:30 and 13:10 today?

we had to change network cabling on the colo server that hosts the munin process . And then after ages of fighting the server we found out at the IPMI card was misbehaving and eating pacakges ...

edit: Been looking more closely at the graphs now, comparing them to ubi's and cheeseboy's. One thing I notice is that my system was much higher on swap before the crash than what seems to be "normal". Any chance this could be relevant?

could be very important, because other users with stability issues also indicated memory shortages
.

[rasmus]

Thanks to both Cheeseboy and Ubi, and a sorry from me that i was "hijacking" your thread with my problems that seems unrelated to the case here

Though, around 30minuttes ago my B3 went down again, this time i was playing around so no heave load, just an idling B3. Blue LED, no connection with SSH (no IP is getting assigned), no respons to pushing the power button.
Looking through the logs no messages at all from aprox. an hour before...

If i can do anything, do tell (all though i'm rather noobish with linux), else i will just keep lurking in this thread and hope you guys find some solution

//Rasmus

[Cheeseboy]

Hi Rasmus,

Sounds like the same issue.
Please email support_AT_excito.com. Your setup might be helpful for them in trying to resolve the issue.
I know they are looking in to this. The problem (not a problem really) for me is that it seems to have stopped crashing (and I have no idea why)...

Cheers,

Cheeseboy

[DanielM]

Another one for me. This time it seems I had not much in swap it seems. This is really frustrating!

/Daniel

edit: And again this morning. Actually, since I installed Munin it has crashed every day for me

edit2: And now another crash. Five days uptime now. Nobody was actively using the network this time. Could the sudden peak in the load average graph just before the crash be related?

[Cheeseboy]

Blam! It happened again, after nearly two months!

I pulled the plug at 15:08, last message in syslog is at 15:02. System is up again at 15:10.
I looked at Ubi's page of Munin stats, and I notice a lot of difference in my system compared to Daniel's and Ubi's in "Individual interrupts".
Particularly "mv.xor.0" and "mv.xor.1", but I have no idea what this means...
Any ideas?

Cheers,

Cheeseboy

[Ubi]

no idea what that means. But if you and Daniel suffer from the same problem, then this seems to be not related. Daniel did not have high mv_xor levels before the crash. He just found an increase in swap (which also did not correlate completely with observed crashed...)

We should get some more bubbae on the list for references though...

[Cheeseboy]

I can only add that by cutting my rather large outgoing traffic for a while, that the mx_xor counters and interrupts in general were directly related to that.

When the traffic stopped, those counters went down.

EDIT: this was traffic routed through the B3, incoming on ETH1, outgoing on ETH0.