conntrackd daemon stops: is conntrackd running?

conntrackd daemon stops: is conntrackd running?

[Pablo Florencio]

[-- Attachment #1: Type: text/plain, Size: 1297 bytes --]

Hi to everyone.

First of all, I apologize for my english.

I have installed conntrackd in my high availability firewall cluster and 
I found one problem that I can't resolve. The installation was ok, I 
used Heartbeat instead of Keepalived for set up the virtual ips and the 
conntrackd script and it seems to work fine- I have checked that both 
nodes can dump the internal and external cache correctly.

The problem comes when tried to use conntrackd a few minutes later since 
the daemon has started, for example, if I want to dump statistics 
collected by daemon I found this error message:

 >conntrackd -s
 >can't connect: is conntrackd running? appropriate permissions?

The commands -c -f -R -k, etc, give the same result. If I try to restart 
the dameon I get this message:

 >conntrackd -d
 >lockfile `/var/lock/conntrack.lock' exists, perhaps conntrackd already 
running?

If I look for the conntrackd process ( ps -ef | grep conntrackd ) I 
found that it has disappear, so I have to delete the .lock file to be 
able to start ir again.

Is there any problem using heartbeat instead of keepalived? I can see it 
working fine for a few minutes but after that I mysteriously found this 
error in both nodes.

Any help or ideas would be very much appreciated.

Regards
Pablo Florencio.



[-- Attachment #2: S/MIME Cryptographic Signature --]
[-- Type: application/x-pkcs7-signature, Size: 3881 bytes --]

Re: conntrackd daemon stops: is conntrackd running?

[Maximilian Wilhelm]

Pablo Florencio scripsit:

Hi!

> I have installed conntrackd in my high availability firewall cluster and 
> I found one problem that I can't resolve. The installation was ok, I 
> used Heartbeat instead of Keepalived for set up the virtual ips and the 
> conntrackd script and it seems to work fine- I have checked that both 
> nodes can dump the internal and external cache correctly.

> The problem comes when tried to use conntrackd a few minutes later since 
> the daemon has started, for example, if I want to dump statistics 
> collected by daemon I found this error message:

[conntrackd dies after some time]
> Is there any problem using heartbeat instead of keepalived? I can see it 
> working fine for a few minutes but after that I mysteriously found this 
> error in both nodes.

I'm running conntrackd with heartbeat wihout a problem for a while now.
Most probably you hit the same bug in libnetfilter-conntrack than I
did some weeks ago.

See http://marc.info/?l=netfilter-devel&m=120942089523985&w=4 for
details.

Ciao
Max
-- 
	Follow the white penguin.

Re: conntrackd daemon stops: is conntrackd running?

[Pablo Neira Ayuso]

Pablo Florencio wrote:
> Hi to everyone.
> 
> First of all, I apologize for my english.

Your English is just fine.

> I have installed conntrackd in my high availability firewall cluster and
> I found one problem that I can't resolve. The installation was ok, I
> used Heartbeat instead of Keepalived for set up the virtual ips and the
> conntrackd script and it seems to work fine- I have checked that both
> nodes can dump the internal and external cache correctly.

This report doesn't help that much, what do the log file says? Please,
also include the version of the software that your using, including the
libraries.

> The problem comes when tried to use conntrackd a few minutes later since
> the daemon has started, for example, if I want to dump statistics
> collected by daemon I found this error message:
> 
>>conntrackd -s
>>can't connect: is conntrackd running? appropriate permissions?
> 
> The commands -c -f -R -k, etc, give the same result. If I try to restart
> the dameon I get this message:
> 
>>conntrackd -d
>>lockfile `/var/lock/conntrack.lock' exists, perhaps conntrackd already
> running?

Apparently, it seems that it is crashing but I need more information to
give you a hand.

> If I look for the conntrackd process ( ps -ef | grep conntrackd ) I
> found that it has disappear, so I have to delete the .lock file to be
> able to start ir again.
> 
> Is there any problem using heartbeat instead of keepalived? I can see it
> working fine for a few minutes but after that I mysteriously found this
> error in both nodes.

Not really, I truly prefer keepalived but it's up to you.

-- 
"Los honestos son inadaptados sociales" -- Les Luthiers