Network unreachable with IPv6, DA resets every x min

neo-hippie

Verified User
Joined
May 27, 2014
Messages
69
Location
The Netherlands (Holland)
Hello,

i have setup IPv6 on my DA server.
but for some reason my default route (output of ip -6 route) is being deleted.

i believe it's when the cron is executing dataskq

also in /etc/network/run/ifstate (symlink) the rule eth0=eth0 constantly is being removed.
but i don't believe that solves the problem, because the network is working with or without it.

but when the network isn't working. i need to restart the server and everything is working again.

can somebody help?
 
Hello,

Yes, here you can find both commerce and free of charge help.

If you need a commerce service please send a PM to those of us who you trust more for details. If you need a free assistance, then we need more details from your server so we could help you. Check directadmin and system logs for more clues.

As far as I know even if you use IPv6, you still need IPv4 on directadmin server to let the things work. Do you have IPv4? And directadmin does not delete any IP with cron. To avoid restarting of your server you might need to get/write a script which whould restart your network or add missing IPs with cron if connectivity gets down.
 
Hi Alex,

i went to the forum for free help :eek:

and basically this is the info i have.
i have 1 IPv4 and a /112 IPv6 range.
both are setup in directadmin and configured in /etc/network/interfaces.

i cannot find anything ... ANYTHING.. which changes after directadmin runs dataskq
i let: ping6 2001:4860:4860::8888 running
and in another terminal i run: tail -f /var/log/syslog

this goes fine for half an hour. (multiple cron dataskq pass) until 1 dataskq does something
and than network is unreachable.

when i do: ip -6 route the output with a working network is:
2001:xxxx:4300:xxxx::/64 dev eth0 proto kernel metric 256
fe80::/64 dev eth0 proto kernel metric 256
default via fe80::2d0:ff:fefc:3000 dev eth0 proto ra metric 1024 expires 1772sec
default via fe80::2d0:ff:fefb:9000 dev eth0 proto ra metric 1024 expires 1772sec

but when network is unreachable it displays:
2001:xxxx:4300:xxxx::/64 dev eth0 proto kernel metric 256
fe80::/64 dev eth0 proto kernel metric 256

the output of ifconfig doesn't change.. as does ip -6 addr
the only thing changing is ip -6 route.

than i reboot the server. and everything works again.
for an half hour...
 
is there a way to create logfiles of what dataskq does?
i have found this forum: http://forum.directadmin.com/showthread.php?t=39742

but it's stil like searching a needle in a haystack for me.

the only thing i get is with dataskq d800 but this is after IPv6 loses connection.:
root priv set: uid:0 gid:0 euid:0 egid:0
pidfile written
staring queue
exim1: Unable to find ip_after from
done queue
Debug mode. Level 800
 
Last edited:
I hardly can imagine that dataskq removes your ips or routes. That must be caused by something else.

dataskq d800


add this into crontak and redirect output into a file.
 
i have done that.

edited the /etc/cron.d/directadmin_cron
changed:
* * * * * root /usr/local/directadmin/dataskq d800 >> /var/log/dataskq.log 2>&1

and the output was for half an hour:
Debug mode. Level 800
root priv set: uid:0 gid:0 euid:0 egid:0
pidfile written
staring queue
done queue

until the network went down, thats when the output changed to:
Debug mode. Level 800
root priv set: uid:0 gid:0 euid:0 egid:0
pidfile written
staring queue
exim1: Unable to find ip_after from
done queue


i do not know what dataskq does. don't know if this causes the route to fail.
i do know that in the system log that is the one and only thing that's executed at the time the network fails.

i have viewed every cronjob.. every log.. and can't find anything.

is it possible that it's an external thing? perhaps the network of my hosting provider has glitches.
and the route doesn't come back up? or something?
but i have another server (without DA) and that one isn't experiencing this issue.
 
Update,

pin6 is now running for almost 1,5 hour. and everything is still up and running.

i have removed pop3s and imaps from dovecot.conf as this gave an error (depreciated)
and restarted dovecot

i had an error in exim mainlog:
2014-07-17 11:15:29 socket bind() to port 25 for address (any IPv6) failed: Address already in use: waiting 30s before trying again (5 more tries)
after the reboot. but this went away by itself.
i have checked if exim is listening on port 25. and it is. so no problems there.

besides that i have done nothing but checking log files.
 
Neither directadmin nor the services it supports has nothing to do with routes; they do not remove IPs and routes. So whatever the reason it has nothing to with Directadmin I'd rather say. Anyway I'm not going to convince you in that if you don't want to believe it.


I don't have enough information to determine what might cause the issue you are facing there. Sorry to say that, but if you want a help you'd better hire somebody (not obligatory me, you may choose anyone here) or even open a ticket with you hosting company.
 
Hi Alex,

it not that i don't want to believe.
absolutly not. i'm open to all suggestions.

it's just that all the evidence points there.
if you can give me pointers to look somewhere else i will provide all the information needed.

i do not need a step by step manual from this forum.
just someone that can point me in the right direction, as IPv6 is new to me.
and well (not to offend anyone) DA isn't that clear in config / debugging to me.

btw. the network is now up for 5 hours without interruption.
i wish i could tell what's changed, and i hope it stays fixed..
 
okey UPDATE time,

i can say for sure that it's not DirectAdmin, as Alex stated.
i also encounter this on another server.
although my DirectAdmin server now runs perfect for roughly 52 hours.

still haven't got a idea on what's cosing this.
there is nothing in the logs, no activity what so ever.
 
Back
Top