Tentacle not starting on reboot

[Sancho]

Like it

Up

0

Down

Drop it

::

Hi,

I have had a problem in both version 4 and version 5 of Pandora FMS where the Tentacle server is not starting up when the system reboots.  I have done all the usual things such as that shown here: http://forums.pandorafms.com/index.php/topic,7596.msg15887.html#msg15887 but to no no affect.  When manually starting it there are no issues, warning or errors – it start fine.

I am running Centos 6.4 in a VMWare environment.  There appears to be no errors in either /var/log/messages or /var/log/pandora (any of the logs) to do with this.  When started manually it operates as expected.

In both installation cases (version 4 and version 5) they were clean installs on clean VMs into new databases.

Any help would be appreciated.  Just let me know what you need.

Cheers

Mike.

Any thoughts

Tentacle uses “pandora” user to spawn the server, check it exists and it’s valid. Usually “pandora” user is created on server install, but could be the case of a problem. Check also startup script and run manually to see if something is wrong.

Execute manually chkconfig tentacle_serverd on to force tentacle service startup and tell if is something is reporting any fail or weird  message.

[mensmaximus]

Like it

Up

0

Down

Drop it

::

This issue is by design on Centos 6.x. The tentacle server is started using sudo. On Centos requiretty is active by default which requires a tty to sudo to another user. Therefore you can start the tentacle server using the init script while you are loged on but it fails at boot time.

A workaround is to disable requiretty for the tentacle binary. Open /etc/sudoers in your favorite editor and add the following line:
Defaults!/usr/bin/tentacle_server !requiretty

This will make sure the tentacle server is starting at boot time.

Imho the tentacle server could be improved in many ways. One would be to impersonate as user pandora from within the script instead of using sudo. Introducing debugging using syslog and an easy pid handling (define tentacle.pid in the init script) would be good to. Furthermore adding ReusePort => 1 to the options while binding the socket would avoid the error message “Cannot open socket for address 0.0.0.0 on port 41121” in case you have IPv6 and IPv4 active.

For the developers in tentacle_server line 522 (sub start_server) change
– ReuseAddr => 1,
+ ReuseAddr => 1,
+ ReusePort => 1

Cheers

Michael