Hola Juanjo,
I think I understand the problem. Look at these processes:
-------------------------------------------------------------------------------
root 4574 0.0 0.0 66984 4384 ? Ss Dec21 0:03
globus-gass-cache-marshal: accepting connections
2449 27792 0.0 0.0 5200 2752 ? S 01:03 0:00 \_
/opt/globus/libexec/globus-gass-cache-util -add -t
https://grid003.ft.uam.es:42985/11577/1261439942/
https://vobox02.pic.es:47619/home/pilotespinal/AnalysisFactory/runpilot3-wrapper.sh
2279 28299 0.0 0.0 5196 2744 ? S 01:05 0:00 \_
/opt/globus/libexec/globus-gass-cache-util -add -t
https://grid003.ft.uam.es:46987/11602/1261439958/
https://vobox02.pic.es:54750/home/prodespinal/PilotFactory/runpilot3-wrapper.sh
2279 28641 0.0 0.0 5196 2748 ? S 01:08 0:00 \_
/opt/globus/libexec/globus-gass-cache-util -add -t
https://grid003.ft.uam.es:40875/11631/1261439961/
https://vobox02.pic.es:54750/home/prodespinal/PilotFactory/runpilot3-wrapper.sh
2279 29114 0.0 0.0 5196 2744 ? S 01:11 0:00 \_
/opt/globus/libexec/globus-gass-cache-util -add -t
https://grid003.ft.uam.es:35047/11634/1261439961/
https://vobox02.pic.es:54750/home/prodespinal/PilotFactory/runpilot3-wrapper.sh
2279 29118 0.0 0.0 5196 2744 ? S 01:11 0:00 \_
/opt/globus/libexec/globus-gass-cache-util -add -t
https://grid003.ft.uam.es:46246/11680/1261439984/
https://vobox02.pic.es:54750/home/prodespinal/PilotFactory/runpilot3-wrapper.sh
-------------------------------------------------------------------------------
The globus-gass-cache-marshal allows up to 5 concurrent processes to
stage files into or out of the CE. Normally each process is short-lived,
but these all hang and cause jobs from _all_ users to suffer.
Hung processes are killed after 5 minutes, which probably is too long;
the "tick" parameter in /opt/globus/etc/globus-gass-cache-marshal.conf
can be adjusted, but we rather need to figure out why the commands hang.
It looks like the PanDA factory on vobox02.pic.es has the wrong value for
the GLOBUS_TCP_PORT_RANGE set! In Condor speak: LOWPORT and HIGHPORT.
Demonstration for a bad port and a good port:
-------------------------------------------------------------------------------
$ telnet vobox02.pic.es 47619
Trying 193.109.175.147...
telnet: connect to address 193.109.175.147: Connection timed out
telnet: Unable to connect to remote host: Connection timed out
-------------------------------------------------------------------------------
$ telnet vobox02.pic.es 20200
Trying 193.109.175.147...
telnet: connect to address 193.109.175.147: Connection refused
telnet: Unable to connect to remote host: Connection refused
-------------------------------------------------------------------------------
The pilot factory is run by Xavier Espinal (CC).
If this is not fixed on the VOBOX and you want your CE to work for others,
you would need to ban Xavier's DN and kill the unwanted processes...
|