Vm slots full but VM are not shut down

[miccloud]

Hi,
These are my cloud scheduler config file:
http://www.megaupload.com/?d=TB51L294
http://www.megaupload.com/?d=G508LH86

I have vm_slots=2.

In my job file I add this lines:
Requirements= VMType =?= my-lab.vm
+VMAMI= "ami-code"

When I submit jobs they are done correctly.

So when condor comlpete these jobs I have the two VM in state
unclaimed and they are not shut down.

If I type cloud_status I have correctly returned this:
user@server:/etc/init.d$ cloud_status -m
VM ID VM TYPE STATUS CLUSTER
i-dd704fb6 default Running aws.amazon.com
i-cd6e51a6 default Running aws.amazon.com

cloud_status
Resource pool Resource Pool:
NAME CLOUD TYPE NETWORK ADDRESS
AmazonEC2 AmazonEC2 aws.amazon.com

What is wrong? Has CloudScheduler to shut down vm in UnClaimed state?

Who can help me?

Thanks a lot.

[miccloud]

This is the cloud_scheduler.log file that describe one cloudscheduler
session :
http://www.megaupload.com/?d=XDB4REWD

Thanks a lot.

[miccloud]

I have installed the last version from git hub and it works.

Does +VMKeepAlive tag work in this version?

Is correct +VMKeepAlive="60" if I want that vm's stay up for 1 hour
after they have finish the job?

I would use it because on Amazon EC2, each AMI costs, is counted for
each hour.

Thanks a lot.

[Michael Paterson]

miccloud wrote:
> I have installed the last version from git hub and it works.
>
> Does +VMKeepAlive tag work in this version?
>
> Is correct +VMKeepAlive="60" if I want that vm's stay up for 1 hour
> after they have finish the job?
>
>

correct.
Normally when the jobs are done and machine becomes Idle and Unclaimed
they are shutdown, if VMKeepAlive is set, when the machines become Idle
it sets the timeout, after the KeepAlive time has passed the machine
will then be shutdown (if still idle)
This was put in since one of the other users was submitting test jobs
that would sometimes fail and require some minor tweaks before
resubmitting, this allows the VM to remain up so he would not have to
wait for it to boot up again.

[miccloud]

Are you sure that currently working this feature?

I have set a job with:

Universe= vanilla
Executable= /usr/local/bin/script.py
Arguments= --json file.json
should_transfer_files= YES
when_to_transfer_output= ON_EXIT_OR_EVICT
transfer_input_files= /tmp/efa89b52-7771-4946-a018-434ce5499597/
file.json,/tmp/efa89b52-7771-4946-a018-434ce5499597/
projection_template.xml
Error= /tmp/efa89b52-7771-4946-a018-434ce5499597/log/
om_projection.err.log.$(PROCESS)
Log= /tmp/efa89b52-7771-4946-a018-434ce5499597/log/job.log
Requirements= VMType =?= my.vm
+VMAMI= "ami-cd0be2a4"
+VMKeepAlive= "60"
Queue

And vm_slots=2;

When I launch this job after it is finished Cloud Scheduler shut down
the machine and don't wait 60 minutes.

Is it correct my job?

Thanks a lot.