TrigMC software release coordination: PTD's logbook - page for November 2018

Notes

Use the shift date in reverse chronological order to enter any new Merge Requests.

For each shift date show the NEW MRs, and any related relevant comments/questions; note down when I (or some one else) merged/closed each request, and mark the MR as done/closed by ticking the box ( [x]). If an MR takes longer than a day to merge, leave it listed under the original shift date, but add comments prefaced with the date of the comment/action. Only MRs that are still active should have the box unticked ( [ ]).

Therefore each MR will only appear once in the log, under the date of the day when I first became aware of it.

30/11/2018

[x] Check nightlies

11.44am Checked both GIT and ATN nightlies: all tests completed, with same overall resuts as yesterday. However there is one compilation error (due to the ART package) where there were none previously. Checked the output and it seems it is the "usual" issue with a duplicate symlink...that is, and error that is not specific to the package, but really just a spurious compilation-time error.

29/11/2018

[x] !16338 Sweeping !16297 from 21.1 to 21.0-TrigMC. Fix for trigger ART tests (opened by Atlas Nightlybuild)

10.22am This was opened overnight automatically, as !16297 was "alsoTargeting:21.0-TrigMC". "Review-approved" by L1 shifter, and CI pipeline successfully concluded. Merged OK.

[x] Check nightlies

10.51am Checked both the GIT and ATN TrigMC nightlies: exactly same overall results as yesterday.

28/11/2018

[x] Check nightlies

11.23am Had a quick look at the nightlies (GIT and ATN): exact same overall test results as the previous nightly (as expected, given that there were no new MRs since previous nightly).

[x] !16299 Merge nightly/21.1/2018-11-27T2139 into the branch 21.0-TrigMC (opened by Rafal Bielski)

11.05am This MR was opened about an hour ago. It is step 2 of the plan to phase out Run-2 trigger releases, to give way to the preparations for the Run-3 releases. It brings TrigMC in line with the 21.1 branch (step 1, see !16228 below, was to bring TrigMC in line with 21.0.). The CI pipeline is listed as "failed"... but based on what happened with !16228, I am guessing it could just be a communication issue (eg along the lines of can't communicate with job server, tehrefore job appears to have "failed" even thought it is running happily along...). (For !16228, results from the supposedly "failed" pipeline job, appeared out of nowhere several hours later... and it then turned out the pipeline had successfully passed all steps.)

3.24pm CI pipeline results have now appeared, as expected: all OK. Merged OK.

27/11/2018

[x] Check nightlies

10.53am Quick check of the nightlies (GIT and ATN): nothing out of the ordinary, but noted that the percentage (%OK) of tests successfully completed has dropped slightly, from 66 to 64.

[x] !16247 updates for Trigger ART tests (opened by Julie Kirk)

10.50am Opened yesterday evening. CI pipeline OK and review-approved 2 hrs ago. Merged OK.

26/11/2018

[x] !16228 Merge release/21.0.90 into the branch !21.0-TrigMC (opened by Rafal Bielski)

10.51am This was opened in the last hour. RB added the "sweep:ignore" label to avoid sweep to master (as presumably all changes were previously swept to master direct from 21.0). CI pipeline job seems to have failed a couple of mins. after it started... haven't managed to establish why.

3.43pm A retry of the build was triggered by Rafal about an hour ago, and the CI pipeline has just concluded without problems. As this MR is based on the latest 21.0 (21.0.90) branch, will merge it without waiting for L1 shifter approval. Merged OK.

11.00am Quickly checked the nightlies on NICOS (GIT and ATN tests): nothing looks out of the ordinary, when compared with preceding nightlies. All tests completed, and identical warning/error counts.

Plan for releases in coming days/weeks

A few days ago (21/11) Rafal and Julie wrote to let us know of the plans for 21.0-TrigMC in the coming days and weeks.

Dear Pedro and Veronique, we would like to let you know in advance of our trigger release plans for the next weeks, which include phasing out the Run-2 branches and moving focus to 21.3 and master. Please see slides 7-8 in our presentation from today's Trigger General Meeting: https://indico.cern.ch/event/775218/contributions/3221988/attachments/1756785/2848617/20181121_ReleasesAndValidation.pdf

In particular, we plan the last big MRs to 21.0-TrigMC in the next two weeks and then doing a merge from 21.0-TrigMC to 21.3. We will still leave 21.0-TrigMC open until the validation of 21.3 is complete, but we will likely decrease the frequency of builds from nightly to weekly over the holidays and in January. We plan to stop 21.0-TrigMC builds when 21.3 is fully validated.

Since we will not have any TrigMC branch over the shutdown, we would like to remind you about other opportunities to gain OTP by contributing to Trigger Release and Validation activities. As previously indicated in an email on 1st November, we have open bookings for weekly validation shifts and monthly rotating validation coordinator. More details are available in slide 5 of the presentation linked above.

Please let us know if you have any questions or concerns.

Cheers, Julie and Rafal

23/11/2018

10.53am Checked the TrigMC nightlies (GIT and ATN): nothing looks out of the ordinary.

[x] !16172 Sweeping 16122 from 21.1 to 21.0-TrigMC. Add art-result code when references are missing (opened by Atlas Nightlybuild)

10.45am MR opened automatically overnight, as !16122 was labelled "alsoTargeting:21.0-TrigMC". Pipeline completed OK, and "review-approved" label added. Merged OK.

21/11/2018

[x] !16119 Updates to ART tests - add regtest files to output (opened by Julie Kirk)

5.35pm Opened earlier today. Pipeline has passed successfully and review-approved label. Merged OK.

[x] !16111 Sweeping !16092 from 21.1 to 21.0-TrigMC. Fix to rootcomp.py command in trigger ART tests (opened by Atlas Nightlybuild)

9.07am This MR is an automatic consequence of !16092 (see below). CI pipeline ran OK. Review-approved label. Merged OK.

20/11/2018

[x] !16092 Fix to rootcomp.py command in trigger ART tests (opened by Julie Kirk)

3.33pm Opened an hour ago. MR is for 21.1, but carries the label "alsoTargeting:21.0-TrigMC". This means I don't need to approve/merge it myself (someone else will), but expect the MR to be applied automatically to 21.0-TrigMC in the 24hrs following merging into 21.1.

16/11/2018

11.53am Checked the nightly on NICOS: all looks good.

[x] !15975 Sweeping 15910 from 21.1 to 21.0-TrigMC. fixes for trigger ART test monitoring (opened by Atlas Nightlybuild)

11.30am MR opened automatically overnight. CI pipeline finished OK. As this is a sweep of a previous MR, it is "safe" to merge, despite review-pending label. Merged OK.

14/11/2018

[x] !15889 Merge branch 'ARTtestError' into '21.1' (opened by Julie Hart Kirk)

10.43am Request opened a few minutes ago. CI pipeline has just started running. Title of MR is slightly weird (?), as it mentions 21.1... not 21.0-TrigMC...?

1.57pm Julie has confirmed that this is a cherry-pick of a previous MR (entitled "Merge branch 'ARTtestError' into '21.1' "), now being applied to 21.0-TrigMC. She has updated the title accordingly, to clarify.

15/11/2018 12.35pm The MR has now been review-approved. Merged OK.

9/11/2018

10.37am Had a quick look at the latest nightly, following on from the two MRs yesterday. (Not available on the ART page: still issues with it...) From the NICOS pages, it looks like there are no differences (as far as building and testing) between the latest nightly and the one form the previous nights.

8/11/2018

[x] !15734 time prints the full cmd line before real..user..sys (for branch 21.0-TrigMC) (opened by Oleg Zenin)

10.40am Request opened yesterday night. The pipeline has concluded OK. A JIRA thread is mentioned: it is long and detailed, and is "watched" by recognised experts (Obreshkov, Undrus); I therefore believe the issue has been ironed out and there has been convergence on a fully worked-out solution. Happy to merge, despite L1-review still pending. Merged OK.

[x] !15726 ATR-19044: bug in Run-1-style taggers in the split configuration (opened by Carlo Varni)

10.26am This MR was opened yesterday evening. The pipeline has passed. However: the review-pending-L1 label was never applied (I have never seen this on an MR before...), so there has been no L1 shifter review. Also: the JIRA ticket does not add anything; it was not watched by any of the usual experts (only the author and one other colleague). The requester states on the MR that it is a simple bug fix with no consequences on other triggers. I have looked at the actual committed changes, and they do indeed look like a simple bug fix, nothing complicated... Therefore will risk Merging it without explicit L1 shifter approval... Merged OK.

-- PedroTeixeiraDias - 31 Oct 2018