Yes, as I said in my one of me previous messages, I was wrong about this,
the job only dies if you explicity send a SIGKILL/INT/TERM to the qrsh
process, so this is a non-issue.
Thanks for your help.
2012/8/28 Reuti
> Am 28.08.2012 um 19:20 schrieb Julien Nicoulaud:
>
> > Yes, exactly that.
>
>
Hello all,
thanks for the advises. I applied the procedure detailed on the wiki and jobs
are now correctly dispatched. Just note that the attributes of the qconf
-msconf must all be changed to match those displayed on the procedure
(queue_sort_method and load_formula but also the others).
Than
I was wondering, how people deal with oom conditions on there cluster.
We constantly have machines that die because the oom killer takes out
critical system services.
Has any experiance with the oom_adj proc value, or a patch to grid to
support it?
/proc/[pid]/oom_adj (since Linux 2.6.11)
I have a mix of high-throughput and long wait jobs. We classify and
prioritize jobs based on runtime. We use a jsv to set
devel # job length < 1hr
short # job length < 6hr
medium # job length < 2day
long # job length < 1wk
xlong # job length > 1wk (goes to ACLed queue)
Users have to spec
Am 29.08.2012 um 17:02 schrieb Ben De Luca:
> I was wondering, how people deal with oom conditions on there cluster.
> We constantly have machines that die because the oom killer takes out
> critical system services.
>
> Has any experiance with the oom_adj proc value, or a patch to grid to
> supp
We use mem_free variable as a consumable. Then, we use a cronjob called
memkiller that terminates jobs if they go over their requested (or
default) memory allocation and
1. Swap space on node is used
2. Swap rate is greater than 100 I/Os per second
The user gets emailed with a report if this
Am 29.08.2012 um 17:21 schrieb Brian Smith:
> We use mem_free variable as a consumable. Then, we use a cronjob called
> memkiller that terminates jobs if they go over their requested (or default)
> memory allocation and
It would be more straight forward to use directly h_vmem. This is controll
On Wed, Aug 29, 2012 at 8:33 AM, Reuti wrote:
> Am 29.08.2012 um 17:21 schrieb Brian Smith:
>
>> We use mem_free variable as a consumable. Then, we use a cronjob called
>> memkiller that terminates jobs if they go over their requested (or default)
>> memory allocation and
>
> It would be more s
We found h_vmem to be highly unpredictable, especially with java-based
applications. Stack settings were screwed up, certain applications
wouldn't launch (segfaults), and hard limits were hard to determine for
things like MPI applications. When your master has to launch 1024 MPI
sub-tasks (qr
Dave,
Thats exiting to know..so if I understand correct.
* I should be able to build a full blown cluster "possibly 400-Nodes" , I
know you tested it till 250.
* If I install the following snapshot..
Eg:-
http://arc.liv.ac.uk/downloads/SGE/snapshots/sge-20120816.tar.gz
Dave,
I didn't have much of a success in installing the
sgeinspect. May be the old instructions are not good enough. I am trying
to do this on a ubuntu based "11.04" GridEngine cluster
Platfrom-RTM is a web-based realtime solutions build on
cacti. Maintains the full history of cluster
Dear All,
I encounter a problem recently.
The version GE2011.11p1 that installed at Linux OS was OK. Then I wanna
add Solaris_64 into the group, but no go.
./aimk -no-java -no-jni -no-secure -spool-classic -no-dump -no-hwloc
-only-depend
Readlink: command not found
Making in SOLARIS64/
Hi Harris,
>From the error message, it looks like you don't have make installed...
Are you able to run "make" from an interactive shell??
Rayson
On Wed, Aug 29, 2012 at 10:08 PM, Harris He, Kun - CD
wrote:
> Dear All,
>
>
>
> I encounter a problem recently.
>
> The version GE2011.11p1 that i
13 matches
Mail list logo