teward | sarnold: yo be alive | 00:13 |
---|---|---|
teward | i need to pick your brain | 00:14 |
sarnold | heya teward | 00:15 |
teward | incoming PM flood | 00:15 |
Aison0 | ahhh :-( bind9 keeps segfaulting and segfaulting, no idea why | 14:34 |
Aison0 | and it creates a very big "core" file | 14:35 |
Aison0 | 800M | 14:35 |
Aison0 | 800M on master | 14:35 |
Aison0 | 3.6G on slave! | 14:35 |
rbasak | dmesg/kern.log will usually give you the immediate reason | 14:35 |
Aison0 | how do I pip stderr to file? | 14:36 |
Aison0 | pipe | 14:37 |
rbasak | your-command 2>filename | 14:37 |
RoyK | Aison0: pastebin output of dmesg -T, or at least the relevant lines | 14:41 |
Aison0 | dmesg shows simply nothing | 14:41 |
Aison0 | but this is the whole bind9 output from start until crash | 14:45 |
Aison0 | https://paste.ubuntu.com/p/vQZqkMxZKN/ | 14:45 |
RoyK | how long does it take? | 14:46 |
Aison0 | 10-15 seconds? | 14:46 |
Aison0 | but it depends | 14:46 |
Aison0 | sometimes longer | 14:46 |
rbasak | I suspect you have filesystem corruption or a hardware fault there | 14:54 |
rbasak | Especially if dmesg output really is blank | 14:54 |
sdeziel | it's not segfaulting but tripping on an assert() | 14:55 |
RoyK | Aison0: I'd start out with monitoring the process with top or htop and look for memory usage. with the sizes of those dumps, it smells like a memory leak | 14:55 |
sdeziel | if those bind servers are caching resolvers, this could explain the size of the core dumps. The assertion is worrying though | 14:56 |
RoyK | still, if it takes so short a time for them to crash, it should be easy to just follow the mem usage for both process and system (and swap, of course) | 14:57 |
RoyK | but then, if that happened, there should be an OOM showing up in dmesg | 14:57 |
Aison0 | RoyK, it really depends. Now it works for several minutes | 14:57 |
Aison0 | it also worked over night | 14:57 |
RoyK | Aison0: how many clients do you have, using that server? | 14:58 |
Aison0 | before my message, it started crashing every few seconds | 14:58 |
RoyK | or those | 14:58 |
Aison0 | RoyK, around 200 | 14:58 |
RoyK | not a lot, then | 14:58 |
Aison0 | no | 14:58 |
Aison0 | it also worked for a long time now | 14:58 |
Aison0 | this setup is not new | 14:59 |
RoyK | has there been a bind update recently? | 14:59 |
Aison0 | rbasak, I don't think it is a hardware problem. It happens on primary and secondary server, which are completely different | 14:59 |
RoyK | agreed - this does *not* smell hardware issues | 15:00 |
Aison0 | for what is this "core" file good for? | 15:06 |
genii | understanding the state of the system at the time the problem occurred | 15:07 |
Darkhunter | Hello, is there a change to use LVM raid during installation? | 15:10 |
RoyK | Aison0: you can run 'gdb bind core.xxx' and then run a backtrace to check where it failed. it'll normally require symbols, though, which may not be there | 15:12 |
RoyK | Aison0: I guess that'll be gdb named core.xxx, though | 15:13 |
teward | *pokes rbasak* got a few minutes? | 15:46 |
Aison0 | are there any ppa with newer versions of bind that I can try? | 15:52 |
lotuspsychje | Aison0: we usually advice to use packages from the repos on ubuntu, specific for your ubuntu version | 15:56 |
lotuspsychje | and/or snaps | 15:56 |
sdeziel | Aison0: https://launchpad.net/~isc/+archive/ubuntu/bind seems reputable enough | 15:58 |
sdeziel | lotuspsychje: I agree with you but in this case, bind9 (9.16.1) is tripping on assert() and ISC upstream fixed a bunch of assertion in later 9.16.X | 15:59 |
lotuspsychje | ah nice, yeah some cases might be useful | 15:59 |
lotuspsychje | nice find sdeziel | 16:00 |
Aison0 | sudo add-apt-repository ppa:isc/bind | 16:02 |
Aison0 | Cannot add PPA: 'ppa:~isc/ubuntu/bind'. | 16:02 |
Aison0 | ERROR: '~isc' user or team does not exist. | 16:02 |
Aison0 | :P | 16:02 |
Aison0 | rofl, bind just crshed again | 16:04 |
Aison0 | that's why it is not working ^'^'^ | 16:04 |
oerheks | make sure you have software-properties-common installed, ppa:isc/bind is working here. | 16:25 |
rbasak | teward: o/ | 16:32 |
=== coconut__ is now known as coconut | ||
=== StathisA_ is now known as StathisA |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!