Bug 630937 - Grid QMF job identification incongruity
Summary: Grid QMF job identification incongruity
Keywords:
Status: CLOSED CURRENTRELEASE
Alias: None
Product: Red Hat Enterprise MRG
Classification: Red Hat
Component: condor-qmf
Version: 1.3
Hardware: All
OS: Linux
low
medium
Target Milestone: 1.3.2
: ---
Assignee: Pete MacKinnon
QA Contact: Martin Kudlej
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2010-09-07 12:37 UTC by Martin Kudlej
Modified: 2011-02-15 13:02 UTC (History)
1 user (show)

Fixed In Version: condor-qmf-7.4.5-0.1
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2011-02-15 13:02:00 UTC
Target Upstream Version:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Red Hat Bugzilla 674290 0 low CLOSED [RFE] Grid QMF job identification incongruity in GetJobAd() 2021-02-22 00:41:40 UTC

Internal Links: 674290

Description Martin Kudlej 2010-09-07 12:37:16 UTC
Description of problem:
1) There is one type of job identification in return object: 
result = session.getObjects(_class="scheduler", _package="com.redhat.grid")[0].SubmitJob(ads)

result.Id == hostname1#32  - SubmitHostname#ClusterId

2) There is another type of identification in job list in GetJobSummaries():
{u'Jobs': {u'1.0': {...}} - ClusterId.ProcId

3) Function GetJobAd() requires job id in form: ClusterId.ProcId

I think it will be good to unify identification of job in QMF API and it should be GlobalJobId.

And also I cannot use directly identification from result.Id in GetJobAd function.

Version-Release number of selected component (if applicable):
qmf-0.7.946106-12.el5
python-qmf-0.7.946106-8.el5
condor-qmf-7.4.4-0.9.el5
qmf-devel-0.7.946106-12.el5
condor-7.4.4-0.9.el5

Comment 1 Matthew Farrellee 2010-09-07 13:29:33 UTC
FYI - (2), the fact that ClusterId.ProcId is present as a key should not be relied upon. It is an artifact of not being able to nest lists in maps (only maps could nest in maps).

Comment 2 Matthew Farrellee 2010-09-15 22:19:27 UTC
(2) is no longer the case, GetJobSummaries as of 7.4.4-0.12 returns a list of maps.

Comment 3 Matthew Farrellee 2010-09-15 22:25:37 UTC
(1) is actually SCHEDD_NAME#ClusterId of created submission.

This can be used to locate the job. A job created with SubmitJob will always have a ProcId of 0. It would not be unreasonable to return SCHEDD_NAME#ClusterId.0, which is useful in a call to GetJobAd.

Comment 4 Pete MacKinnon 2010-11-03 16:20:27 UTC
part of FH sha e7ea72c

now we return schedd_name#cluster.proc...makes it easier for chaining test scripts after original submit (hold, release, getjobad, etc.)

Comment 5 Matthew Farrellee 2010-11-18 19:19:53 UTC
Set to be built post 7.4.4-0.17

Comment 6 Martin Kudlej 2011-01-31 10:40:27 UTC
I've tested it on RHEL 4.9/5.6 x i386/x86_64 with:
ruby-qmf-0.7.946106-27.el5
python-qmf-0.7.946106-14.el5
condor-wallaby-tools-3.9-2.el5
qmf-0.7.946106-27.el5
python-condorutils-1.4-6.el5
condor-7.4.5-0.7.el5
condor-wallaby-client-3.9-2.el5
condor-job-hooks-1.4-6.el5
condor-low-latency-1.1-2.el5
condor-wallaby-base-db-1.5-2.el5
condor-qmf-7.4.5-0.7.el5

and 1) and 2) are ok.

3) Is that right that hold, release, getjobad, etc. still need ClusterId.ProcId instead of schedd_name#cluster.proc?

Comment 7 Pete MacKinnon 2011-01-31 15:25:28 UTC
It is for now. I suppose we could make it more consistent. The thing is if you are holding or releasing a job you have already identified (and connected to) the scheduler agent for that job.

Comment 8 Matthew Farrellee 2011-01-31 16:06:52 UTC
Please verify without (3). (3) could be an RFE for further discussion, but does not have a good outlook.

Comment 9 Martin Kudlej 2011-02-01 10:28:38 UTC
3) I've created RFE bug 674290

1) 2) are ok so --> VERIFIED


Note You need to log in before you can comment on or make changes to this bug.