Bug 517124 – sys-cluster/torque-4.1.7 - bad permission on /var/spool/torque/spool/ -//- emerge --config runs pbs_server which expects a running trqauthd -//- --config fails to stop pbs_server on error -//- and so on

• Status: RESOLVED WONTFIX
• Product: Gentoo Linux

[Description Martin Mokrejš 2014-07-14 21:30:31 UTC]

I was repeatedly failing to get "emerge --config torque" working.


So I found the docs (the URL in ebuild does not work anymore).

http://docs.adaptivecomputing.com/torque/torqueAdminGuide-4.5.0.pdf

# pbs_server -t create

But, it turned out per /var/spool/torque/server_logs/20140714 file that:

07/14/2014 23:03:26;0001;PBS_Server.17893;Svr;PBS_Server;LOG_ERROR::Permission denied (13) in chk_file_sec, Security violation with "/var/spool/torque/spool/" - /var/spool/torque/spool/ cannot be accessed
07/14/2014 23:03:26;0001;PBS_Server.17893;Svr;PBS_Server;LOG_ERROR::PBS_Server, pbsd_init failed


So, the ebuild did not setup properly the directory permissions.

# chmod a+rwxt /var/spool/torque/spool/

Let's retry now to get the installation initialized:

#  emerge --config torque

Configuring pkg...

 * Configuring Torque ...
 * Using /var/spool/torque as the pbs homedir
 * Using vostro.doma as the pbs_server
 * Previous Torque configuration detected.  Press Enter to
 * continue or Control-C to abort now

socket_connect error (VERIFY THAT trqauthd IS RUNNING)
Error in connection to trqauthd (15096)-[could not connect to unix socket /tmp/trqauthd-unix: 111]


Doh, I am supposed to do:

# /etc/init.d/trqauthd start
hostname: vostro.bar
pbs_server port is: 15001
trqauthd daemonized - port 15005
#

# emerge --config torque


Configuring pkg...

 * Configuring Torque ...
 * Using /var/spool/torque as the pbs homedir
 * Using vostro.doma as the pbs_server
 * Previous Torque configuration detected.  Press Enter to
 * continue or Control-C to abort now

pbs_server port already bound: Address already in use
 * Failed to start pbs_server                                                                                                                                                                                                                                                                                           [ !! ]

#


Damn, couldn't the einfo() mention it is going to run "pbs_server -t create" on its own? ;-)

# killall pbs_server
#

# emerge --config torque

Configuring pkg...

 * Configuring Torque ...
 * Using /var/spool/torque as the pbs homedir
 * Using vostro.doma as the pbs_server
 * Previous Torque configuration detected.  Press Enter to
 * continue or Control-C to abort now
                                                                                                                                                                                                                                                                                                                       [ ok ]
#

# /etc/init.d/pbs_server start
 * Starting Torque pbs_server ...                                                                                                                                                                                                                                                                                        [ ok ]
# /etc/init.d/pbs_sched start 
 * Starting Torque pbs_sched ...                                                                                                                                                                                                                                                                                         [ ok ]
# /etc/init.d/pbs_mom start  
 * Starting Torque pbs_mom ...
pbs_mom: LOG_ERROR::Permission denied (13) in chk_file_sec, Security violation with "/var/spool/torque/undelivered/" - /var/spool/torque/undelivered/ cannot be accessed
 * start-stop-daemon: failed to start `/usr/sbin/pbs_mom'                                                                                                                                                                                                                                                                [ !! ]
 * ERROR: pbs_mom failed to start
# chmod a+rwxt /var/spool/torque/undelivered
#


OK, finally it works. Please fix the ebuild.

Maybe the permission issues is related to http://www.clusterresources.com/bugzilla/show_bug.cgi?id=200 but I did not really study the patches.

[Comment 1 Justin Bronder (RETIRED) 2015-08-05 17:23:41 UTC]

+*torque-4.2.9-r4 (05 Aug 2015)
+
+  05 Aug 2015; Justin Bronder <[e-mail]> +torque-4.2.9-r4.ebuild:
+  Drop pkg_config() as handling the multiple authentication mechanisms is no
+  longer feasible from within the ebuild (#517124)


Granted, that's probably not the fix you were looking for, but I don't think the ebuild should be wrangling more than one daemon.  There's also multiple authentication schemes which would require different daemons.

I did fix the URL for upstream docs.