[GE users] SGE6 does not backfill

Juha Jäykkä juhaj at iki.fi
Wed Apr 13 10:14:24 BST 2005


    [ The following text is in the "ISO-8859-15" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]

> Well, would be nice, if you could verify the fix. To test it, you would
> need to update all binaries. u3 will not work together with u4. The database did
> not change.

Well... it looks like backfilling works; at least short serial jobs go
right through the queue if there is only one big parallel job waiting for
execution and a suitably long job being run, which prevent the execution
of the big parallel job. Furthermore, if the serial job in the queue
requests too much h_rt, qstat shows

(-l h_rt=36600) cannot run in queue instance "all.q at compute-0-8.local" because it offers only qc:slots=0.000000

for all queues the big job reserves. This is very nice.

Another problem surfaced, though: 

The parallel jobs NEVER run! They transfer to the exec hosts fine (go from
state "qw" to state "t"), but the vanish without leaving a trace ANYWHERE!
I can never see them in state "r". What's up here? Is some change in my
config required?

-- 
		 -----------------------------------------------
		| Juha Jäykkä, juolja at utu.fi			|
		| home: http://www.utu.fi/~juolja/		|
		 -----------------------------------------------

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe at gridengine.sunsource.net
For additional commands, e-mail: users-help at gridengine.sunsource.net




More information about the gridengine-users mailing list