Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Daniel]

From time to time when i reboot Anope or the box, anope loose registrations.
It differs from time how much and what is lost, so it reverts an unknown amount of time in the db.
I've observed this on both 1.8.* and 1.9.* (i thought upgrading would fix the problem).

I've been over both unrealircd.conf, and services.conf trying to find some misconfiguration that i've made?
But both is similar to a config i've used on CentOS & Slackware, and neither one of those installations loses anything when rebooted...

I tried manually runing /os update multiple times before reboots - without any success...

I've come to the point where i'm totally clueless about what's wrong.

I've observed this behavior on a lot of different versions of Ubuntu, Anope, and UnrealIRCd. But ONLY on Ubuntu machines.
Both when using MySQL and the Flatfile DB's for storing the registrations.
And the backup is of no use either - the problem still remains.

Anyone got a clue about what is wrong??

[Adam]

If you can get debug logs of your /os updating and restarting and have it show what you did didn't save, that would help. Also be sure that the databases change on /os update so you know Anope has write privileges.

[Daniel]

I don't got access to one of the machines atm, due to that i haven't got a hold of the sysadmin so he can change IP in the firewall.

The other machine got u+rw on anope.db.
However, when doing 'tail' on the log files, i located that services failed to save everything and it got cut off before getting to the OS STATS at the end of the file, it stopped somewhere in the middle.

I've been googling this for awhile, and it was my last resort to see if anyone else had the same problem (and possibly a solution for it).
I'll get the debug-logs as soon as i observe this problem again (i can't say when it happens next time, but gonna try and force it to happen).

[katsklaw]

Lets throw insuring you haven't reached/exceeded your quota to the mix too. I use ubuntu but not mysql and have no issues at all. Let's also try disabling mysql and see if you can save to flatfile.

[Daniel]

Lets throw insuring you haven't reached/exceeded your quota to the mix too. I use ubuntu but not mysql and have no issues at all. Let's also try disabling mysql and see if you can save to flatfile.

The server i got access to have ~140GB free, and hasn't quota enabled on users.
The server i don't have access to atm have something similar free, and doesn't either have quota enabled.

Services runs under the same account as the ircd, and both servers are low usage (so the file descriptor limit isn't reached).

The problem is there, doesn't matter if i run it with mysql or flatfile, it still happens from time to time.
And as i mentioned in my last post, a quick tail on the backup *.db files showed that one or two of 'em seemed to have been cutoff somewhere in the middle (they didn't get down to the OS STATS at the bottom that i observed in all the others *.db) and lacked a \n on the row it got cutoff also.

[Charles Kingsley]

This is the first time I've heard of this, and to hear it happening on both 1.8 and 1.9 really points at the box unless there's something freakish happening.

Rules out modules if it's on both 1.8 and 1.9 really.

When you do /OS UPDATE do you see the timestamps / size change on the dbs?

[Jan Milants]

Rules out modules if it's on both 1.8 and 1.9 really.

tbh, if it happens on both it rules out almost anything on anopes side since 1.9 has almost no old code left... only thing left is an environmental issue.

And what version of anope are you currently using?
This is posted in the 1.8 forums, but I get the impression you are currently using 1.9..

[Daniel]

When you do /OS UPDATE do you see the timestamps / size change on the dbs?

Yes, both size & time gets updated - when it doesn't forget information that is, when it forgets only time gets updated.

tbh, if it happens on both it rules out almost anything on anopes side since 1.9 has almost no old code left... only thing left is an environmental issue.

And what version of anope are you currently using?
This is posted in the 1.8 forums, but I get the impression you are currently using 1.9..

That was interesting, it got me wondering about a thing. Is Anope sensitive if it's in a environment with heavey I/O usage? Both machines runs services that from time to time use a lot of I/O.

Both 1.8 & 1.9 - one box with 1.8 and one with 1.9.
The one with 1.8 is the one firewalled that i'm waiting for access to.

[katsklaw]

I'm wondering if running inside gdb at debug level 4 would help. Perhaps at least get us to an exit point?