Description of problem: The Longest Running Grid Submissions list reports Duration based on when the QMF Submission object was last created. Not based on the time when the Submission was created, or some other useful metric such as the longest running job within the submission. Fixing this may require extra information from the QMF plugins.
Maybe this part of the default persona overview page should just be removed. I believe it is the only place in cumin where we talk about "longest running". It's not part of the grid persona, and we don't have the correct data anyway.
If you do that, file a BZ to bring it back when we do have appropriate data. The use case of identifying long running submissions is a powerful one.
Changing priority to low based on persona views (only visible in default, not grid or messaging)
More feedback from UI review. Assuming better data becomes available from condor, we should consider putting an "accumulated runtime" or "queue date" column (depends on what info is available) on the submissions table and make it sortable. Adding a BZ for Grid to add data to the schema for submissions.
How we can test this bug? Is there any repro scenario?
Martin, Here at least is a reproducer to show the problem. More about verifying the solution when it is finished. 1. Set "persona: default" in the [web] section in cumin.config 2. Start cumin (and condor if it is not already running) 3. Wait for submissions to show up under Longest Running Submission on the front page. The duration here will be the elapsed time since the QMF submission object was created. 4. Shut condor down and wait for all the submissions to disappear from Cumin's Longest Running Submission display. 5. Start condor again. The durations under longest running Submissions should reset and begin near zero. 6. You can also look in the cumin database to verify that the durations reported match the _qmf_create_time column. $psql -d cumin -U cumin -h localhost cumin=# select * from "com.redhat.grid"."Submission"; (maybe limit the query to a particular submission, etc.
Fixed in revision 5031. Longest Running Submissions on the default persona front page now uses QDate. Schema has been updated to include QDate (includes upgrade script). "Enqueued" column replaces "Scheduler" column on Admin->Grid->Submissions tab. Scheduler link added to submission summary (Admin->Grid->Submissions->pick one) to make up for the lost column. This was done because adding "enqueued" to the submission table made the table too wide with the scheduler column still there.
Technical note added. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. New Contents: Cause Cumin's idea of the age of a submission was based on the creation time of the QMF object representing the submission in the MRG Messaging space because there was not adequate data to determine submission age. Consequence Displays such as the Longest Running Grid Submissions table in the default persona were affected by events in the MRG Messaging space and were not accurate. Fix New data generated by condor and integrated in Cumin gives the earliest queue date of any job included in a submission. Result Cumin uses queue date values as the age of a submission. The Longest Running Grid Submission display should be accurate. Additionally, a new column showing queue date has been added to submission table displays.
Reproduction on RHEL5/6 i386/x86_64: 0. Install cumin and condor: # rpm -q condor condor-7.6.3-0.3.el5 condor-7.6.3-0.3.el6 # rpm -q cumin cumin-0.1.4916-1.el5 cumin-0.1.4916-1.el6 1. Set "persona: default" in the [web] section in cumin.config 2a. Start cumin (and condor if it is not already running) 2b. Submit jobs: # su -c 'echo -e "cmd=/bin/true\nhold=true\nqueue 10" | condor_submit' test 3. Wait for submissions to show up under Longest Running Submission on the front page. The duration here will be the elapsed time since the QMF submission object was created. 4. Shut condor down and wait for all the submissions to disappear from Cumin's Longest Running Submission display. 5. Start condor again. The durations under longest running Submissions should reset and begin near zero. 6. You can also look in the cumin database to verify that the durations reported match the _qmf_create_time column. $psql -d cumin -U cumin -h localhost cumin=# select * from "com.redhat.grid"."Submission"; Verification on RHEL5/6 i386/x86_64: 0. Update cumin and condor: # rpm -q condor condor-7.6.4-0.8.el5 condor-7.6.4-0.8.el6 # rpm -q cumin cumin-0.1.5068-1.el5 cumin-0.1.5068-1.el6 # cumin-admin upgrade-schema 1. skip 2a. Restart cumin 2b. skip 3. Wait for submissions to show up under Longest Running Submission on the front page. 4. Shut condor down and wait for all the submissions to disappear from Cumin's Longest Running Submission display. 5. Start condor again. The durations under longest running Submissions should NOT reset and shoud NOT begin near zero. 6. You can also look in the cumin database to verify that the durations reported now match the new QDate column. $psql -d cumin -U cumin -h localhost cumin=# select * from "com.redhat.grid"."Submission"; 7. Longest Running Submissions on the default persona front page now uses QDate. Schema has been updated to include QDate (includes upgrade script). "Enqueued" column replaces "Scheduler" column on Admin->Grid->Submissions tab. 8. Scheduler link added to submission summary (Admin->Grid->Submissions->pick one) to make up for the lost column. This was done because adding "enqueued" to the submission table made the table too wide with the scheduler column still there. ---> VERIFIED
Technical note updated. If any revisions are required, please edit the "Technical Notes" field accordingly. All revisions will be proofread by the Engineering Content Services team. Diffed Contents: @@ -1,11 +1 @@ -Cause +Previously, Cumin based the age of a submission around the creation times of the QMF object that represented the submission in the MRG Messaging space. However, displays such as the Longest Running Grid Submissions table in the default persona were affected by events in the MRG Messaging space, and could therefore be inaccurate. This update ensures that the data generated by condor and integrated into Cumin present the earliest queue date of any job included in a submission, with the result that the Longest Running Grid Submissions display should now be accurate. In addition, a new column which shows the queue date has been added to the table.- Cumin's idea of the age of a submission was based on the creation time of the QMF object representing the submission in the MRG Messaging space because there was not adequate data to determine submission age. - -Consequence - Displays such as the Longest Running Grid Submissions table in the default persona were affected by events in the MRG Messaging space and were not accurate. - -Fix - New data generated by condor and integrated in Cumin gives the earliest queue date of any job included in a submission. - -Result - Cumin uses queue date values as the age of a submission. The Longest Running Grid Submission display should be accurate. Additionally, a new column showing queue date has been added to submission table displays.
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. http://rhn.redhat.com/errata/RHEA-2012-0045.html