Hello there,
I've a strange error I can't understand.
My network has 2 servers, and when some people (on serv1) change their nick, the "old" nick stay connected on one of the server.
On server 1 (where the user is connected):
/whois Guest
16:29:02 Zeolia -- Guest: No such nick/channel
16:29:02 Zeolia -- [Guest] End of /WHOIS list.
/whowas Guest
16:37:13 Zeolia -- [Guest] ([email protected]) was NewNick
16:37:13 Zeolia -- [Guest] server1.zeolia.net (Sun Apr 29 23:53:19 2012)
16:37:13 Zeolia -- [Guest] End of WHOWAS
On server 2:
/whois Guest
16:49:22 Zeolia -- | [Guest] ([email protected]): NewNick
16:49:22 Zeolia -- | [Guest] #Quizz
16:49:22 Zeolia -- | [Guest] server1.zeolia.net (Serveur 1 @ zeolia.net)
16:49:22 Zeolia -- | [Guest] End of /WHOIS list.
/whowas Guest
16:49:31 Zeolia -- | Guest: There was no such nickname
16:49:31 Zeolia -- | [Guest] End of WHOWAS
Any idea about this strange thing ?
User here... but not here
Re: User here... but not here
That's called a desync, it most commonly happens when ircds have different timestamps or are experiencing heavy latency. Try "/tsctl alltime" and paste the results. IRC is extremely time sensitive, I can't stress that enough. ALL servers should have ntp installed, no exceptions. If you have root, install it, if you don't have root then tell your host to install it if it isn't already.
Configuration - https://www.unrealircd.org/docs/Configuration
FAQ - https://www.unrealircd.org/docs/FAQ
FAQ - https://www.unrealircd.org/docs/FAQ
Re: User here... but not here
Sorry, another idea ?21:21:13 Zeolia -- server1.zeolia.net (server1.zeolia.net): *** Server=server1.zeolia.net TStime=1335813673 time()=1335813673 TSoffset=0
21:21:13 Zeolia -- server2.zeolia.net (server2.zeolia.net): *** Server=server2.zeolia.net TStime=1335813673 time()=1335813673 TSoffset=0
I wonder it's a missed configuration, but "server2" is exactly as "server1", I'd just invert some IPs.
Re: User here... but not here
it's not a config setting. You might want to look at the latency between the servers. You can also try squit'ing the server that doesn't see the user and then connect them again to force a burst.
Configuration - https://www.unrealircd.org/docs/Configuration
FAQ - https://www.unrealircd.org/docs/FAQ
FAQ - https://www.unrealircd.org/docs/FAQ
Re: User here... but not here
I'll try next time.
It's the first time I have this kind of trouble, really strange, isn't it ?
It's the first time I have this kind of trouble, really strange, isn't it ?
Re: User here... but not here
No, it's really not all that strange. Desyncs and ghost users happen at times and more often on nets that aren't taking all the precautions to prevent them. It's because IRC is so time sensitive.
Some things that cause ghosts and desyncs:
1> Not using ntp on all servers.
2> Poor or unreliable bandwidth/providers
3> Poor or unreliable hardware
4> Poor or unreliable internet connectivity between servers and users.
Just to name a few. #4 cannot be controlled except to make sure your servers are linked optimally and there will still be good days and bad days for the things you can't control. #4 can also explain how some users have connectivity issues and others don't.
Some things that cause ghosts and desyncs:
1> Not using ntp on all servers.
2> Poor or unreliable bandwidth/providers
3> Poor or unreliable hardware
4> Poor or unreliable internet connectivity between servers and users.
Just to name a few. #4 cannot be controlled except to make sure your servers are linked optimally and there will still be good days and bad days for the things you can't control. #4 can also explain how some users have connectivity issues and others don't.
Configuration - https://www.unrealircd.org/docs/Configuration
FAQ - https://www.unrealircd.org/docs/FAQ
FAQ - https://www.unrealircd.org/docs/FAQ
Re: User here... but not here
Hello there !
I've checked the two servers: ntp is used every hours on each server, using crontab: 00 * * * * /usr/sbin/ntpdate pool.ntp.org > /dev/null 2>&1
The servers are in datacenters with a pretty good bandwidth and their interconnection is really good.
So, I think the trouble come from an overload of one of the servers, I'll try to analysis their usage. One of them is using pisg every hours, and one of the channel has stats from more than 550 days... Seems to be the reason
I've checked the two servers: ntp is used every hours on each server, using crontab: 00 * * * * /usr/sbin/ntpdate pool.ntp.org > /dev/null 2>&1
The servers are in datacenters with a pretty good bandwidth and their interconnection is really good.
So, I think the trouble come from an overload of one of the servers, I'll try to analysis their usage. One of them is using pisg every hours, and one of the channel has stats from more than 550 days... Seems to be the reason