Package: gridengine-client
Version: 6.2u5-1
Severity: normal
Hi,
I've got an SGE installation with two queues:
wferi@n0:~$ qstat -g c
CLUSTER Q CQLOAD USED RES AVAIL TOTAL aoACDS cdsuE
--------------------------------------------------------------------------------
alma -NA- 0 0 0 4 0 4
szie-aotk 0.00 0 0 5 7 0 2
The above report can be broken up by queue name as expected:
wferi@n0:~$ qstat -g c -q alma
CLUSTER Q CQLOAD USED RES AVAIL TOTAL aoACDS cdsuE
--------------------------------------------------------------------------------
alma -NA- 0 0 0 4 0 4
wferi@n0:~$ qstat -g c -q szie-aotk
CLUSTER Q CQLOAD USED RES AVAIL TOTAL aoACDS cdsuE
--------------------------------------------------------------------------------
szie-aotk 0.00 0 0 5 7 0 2
Now I add a third queue:
wferi@n0:~$ qconf -Aq /tmp/teszt.q
wferi@n0.niif.grid added "teszt" to cluster queue list
wferi@n0:~$ qstat -g c
CLUSTER Q CQLOAD USED RES AVAIL TOTAL aoACDS cdsuE
--------------------------------------------------------------------------------
alma -NA- 0 0 0 4 0 4
szie-aotk 0.00 0 0 5 7 0 2
teszt -NA- 0 0 0 0 0 0
So far so good. But separate reporting doesn't work properly:
wferi@n0:~$ qstat -g c -q teszt
wferi@n0:~$ qstat -g c -q alma
CLUSTER Q CQLOAD USED RES AVAIL TOTAL aoACDS cdsuE
--------------------------------------------------------------------------------
alma -NA- 0 0 0 4 0 4
teszt -NA- 0 0 0 0 0 0
wferi@n0:~$ qstat -g c -q szie-aotk
CLUSTER Q CQLOAD USED RES AVAIL TOTAL aoACDS cdsuE
--------------------------------------------------------------------------------
szie-aotk 0.00 0 0 5 7 0 2
teszt -NA- 0 0 0 0 0 0
You don't need two working queue to exhibit the problem, one is enough,
but I found the demonstration more convincing with two. Here's my queue
configuration (they differ in their first two lines only):
wferi@n0:~$ qconf -shgrpl
@alma
@szie-aotk
@teszt
wferi@n0:~$ cat /tmp/teszt.q
qname teszt
hostlist @teszt
seq_no 0
load_thresholds np_load_avg=1.75
suspend_thresholds NONE
nsuspend 1
suspend_interval 00:05:00
priority 0
min_cpu_interval 00:05:00
processors UNDEFINED
qtype BATCH INTERACTIVE
ckpt_list NONE
pe_list make
rerun FALSE
slots 1
tmpdir /tmp
shell /bin/bash
prolog NONE
epilog NONE
shell_start_mode unix_behavior
starter_method NONE
suspend_method NONE
resume_method NONE
terminate_method NONE
notify 00:00:60
owner_list NONE
user_lists NONE
xuser_lists NONE
subordinate_list NONE
complex_values NONE
projects NONE
xprojects NONE
calendar NONE
initial_state default
s_rt INFINITY
h_rt INFINITY
s_cpu INFINITY
h_cpu INFINITY
s_fsize INFINITY
h_fsize INFINITY
s_data INFINITY
h_data INFINITY
s_stack INFINITY
h_stack INFINITY
s_core INFINITY
h_core INFINITY
s_rss INFINITY
h_rss INFINITY
s_vmem INFINITY
h_vmem INFINITY
This problem disappears as soon as I add a (resolvable) host to the @teszt
hostgroup: the qstat -g c -q <name> outputs become reasonable again,
until I remove the host not belonging there (these queues are dynamic, I
typically can't choose a proper host for the new queue at creation time).
If this problem is too hard to fix, I'll work around it, but first I hope
for a real solution, either in qstat or in my configuration.
Can you offer one?