HELP Please - Domains stop working

jeroman8

Well-Known Member
Mar 14, 2003
410
0
166
For the third time the last 14 days ALL domains on the server stop responding.
Status say that all services is up and running, httpd, Bind, mysql etc.

The only thing that seems to fix it is a Reboot !!

Using the IP's to access the sites works great but NOT the "tilde" feature.
http://IP/~username - NOT working.

Today when this happened a use complained about mysql error just one minute
before all the domains stop responding.
"Can't connect .... socket '/var/lib/mysql/mysql.sock' (2)"

If this can have anything to do with it - I don't know but same thing happened
the last time. Mysql error just before the domain issues.

When it happens:

This is OK:
http://ipx.ipx.ipx.ip - goes to the website

This is Not ok:
http://ipx.ipx.ipx.ip /~username (should work)
http://www.domain.com (all domains not working)

Cpanel support told me to do the following which I
have done but it didn't help:
mv /usr/local/cpanel/cpanel /usr/local/cpanel/cpanel.old
/scripts/updatenow
/scripts/upcp --force
service cpanel restart

So Please - do you guys have any ideas what this can be and how to fix it ?
 

rpmws

Well-Known Member
Aug 14, 2001
1,798
9
318
back woods of NC, USA
jeroman8 said:
For the third time the last 14 days ALL domains on the server stop responding.
Status say that all services is up and running, httpd, Bind, mysql etc.

The only thing that seems to fix it is a Reboot !!

Using the IP's to access the sites works great but NOT the "tilde" feature.
http://IP/~username - NOT working.

Today when this happened a use complained about mysql error just one minute
before all the domains stop responding.
"Can't connect .... socket '/var/lib/mysql/mysql.sock' (2)"

If this can have anything to do with it - I don't know but same thing happened
the last time. Mysql error just before the domain issues.

When it happens:

This is OK:
http://ipx.ipx.ipx.ip - goes to the website

This is Not ok:
http://ipx.ipx.ipx.ip /~username (should work)
http://www.domain.com (all domains not working)

Cpanel support told me to do the following which I
have done but it didn't help:
mv /usr/local/cpanel/cpanel /usr/local/cpanel/cpanel.old
/scripts/updatenow
/scripts/upcp --force
service cpanel restart

So Please - do you guys have any ideas what this can be and how to fix it ?
how much load is on the server? are you over your max connections? have you run tests to make sure your DNS is resolving your domains to your IP? It sounds like a dns issue since IP and IP/~user/ is working ..it shouldn't be apache I wouldn't think. If you restart bind what happens?
 

jeroman8

Well-Known Member
Mar 14, 2003
410
0
166
rpmws said:
how much load is on the server? are you over your max connections? have you run tests to make sure your DNS is resolving your domains to your IP? It sounds like a dns issue since IP and IP/~user/ is working ..it shouldn't be apache I wouldn't think. If you restart bind what happens?
NOTE: IP/~user/ is NOT working
And cpanel said it has nothing to do with BIND.

Load is not high at all and can't have anything to do with it.

Well everything works great at the moment but how should i make these
tests when the problem comes back in a few days ?
Can I just use dnsstuff.com to run these test or what do you mean ?

Thanks
 

sawbuck

Well-Known Member
Jan 18, 2004
1,365
10
168
cPanel Access Level
Root Administrator
Assume you have turned on the "mod_userdir " tweak in WHM > Server Setup > Tweak Security?
The DNS for webbhotellinfo.se looks okay. Is that one of the domains that fails?
Have you checked the httpd.conf "ServerAlias" entries for "some-domain.com" and "www.some-domain.com?
The mysql involvement in the domain failure should be investigated. Have you made changes to /etc/my.cnf?
What OS and version of cPanel/WHM?
 

jeroman8

Well-Known Member
Mar 14, 2003
410
0
166
After a reboot eveything is fixed.
However I have noticed that BIND is not running when this happens
and it's impossible to restart it.
I need to reboot and then it's ok.

This happens with 4-6 days in between and I think it has something to do
with entries being written to the files when someone is adding stuff to be
included in named/bind.

So my BIND/Named (is it the same thing or what ?) if a little fucked up.
Have done upcp --force and also removed the cpanel file and then upcp --force
but it didn't help.

In the middle of this the server wouldn't come back after reboot.
Then mysql started to act really wierd.
Now IMAP is not working . can't fix it so also HORDE and Squirremail is down.

What to do ? - I tell you:
Move all customers first and then burn the server :)
 

MPCN_Russ1

Member
Jun 26, 2003
19
0
151
Hello,

I've had many issues in the past with bind... After you try and restart bind, it may fail due to a permissions issue with named.conf in /etc. The way to resolve this is chmod named.conf to 0700 and chown named.conf to named instead of root.
Example:
chmod 0700 /etc/named.conf
chown named:named /etc/named.conf

When Bind goes down... All services go down... including mysql!

Good luck... Be sure to check back and let us know how that worked.

Thanks,
Russ
 

jeroman8

Well-Known Member
Mar 14, 2003
410
0
166
What is happening now is that now and then BIND stops.

Doing this helps:

service named stop
/scripts/fixndc
/scripts/fixnamed
service named start
/scripts/restartsrv_named
/scripts/restartsrv_bind

when doin all of that BIND is back in biz.
But why does it stop in the first place ?
Why can't I just run service named start if it has stoped ?

I have moved about 100 clients from this server and have
100 more to move before I kill this haunted server.
 

MPCN_Russ1

Member
Jun 26, 2003
19
0
151
What I've come to notice is when you add a new account... If the named.conf file is re-owned by root... it won't do the modification and in turn won't update bind for the new account's domain so that domain will NOT setup... I modified SIM from Rfx to reset the privs on the named.conf when bind appears to be down... now the only thing todo would be to get the add account script todo the same after adding the account... But I haven't yet looked into that because I usually remember to just restart bind myself and let it fail... then in a minute or less sim catches it and restarts it... :) I'm not too active with webhosting... So I only add accounts every now and then.

I wish I had the need to just create my own spawn of cpanel so I can actually control all of the functions like the one's that you can't touch in WHM because they're integrated :( I hope cPanel will just release them in the future... But that's highly doubtable... It's been 10 versions now.

Thanks,
Russ