-
Notifications
You must be signed in to change notification settings - Fork 4.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
DQM: new DQMFileSaver #28588
DQM: new DQMFileSaver #28588
Conversation
Using commit 70fc23c from andrius-k:dqm-new-dqmstore-on-CMSSW_11_0_0_pre5. Includes changes to make the coe compile and run with the old DQMStore.
It looks like this was never actually used in online DQM or HLT in run2. There are plenty of threading-related code paths, but DQMServices/FileIO seems to have taken over at some point.
The code-checks are being triggered in jenkins. |
please test I did remember to run code-format, this time! |
+code-checks Logs: https://cmssdt.cern.ch/SDT/code-checks/cms-sw-PR-28588/13085
|
The tests are being triggered in jenkins. |
A new Pull Request was created by @schneiml (Marcel Schneider) for master. It involves the following packages: DQMServices/Components @andrius-k, @kmaeshima, @schneiml, @cmsbuild, @jfernan2, @fioriNTU can you please review it and eventually sign? Thanks. cms-bot commands are listed here |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
I am surprised that no tests failed, even though #28407 (comment) indicated that the old The comparison looks much more messy than expected though, we'll need to fix a bunch of things it seems. Edit: I see
I don't see what the DQMHistoSizes comparison is doing here though, I assume it is just confused by something. I don't see 100's of KB of new histograms in the bin-to-bin comparison. |
Pull request #28588 was updated. @andrius-k, @kmaeshima, @schneiml, @cmsbuild, @jfernan2, @fioriNTU can you please check and sign again. |
@Martin-Grunewald , thanks, I have alerted FOG and we can see if anybody is around to do a quick hilton test |
@Sam-Harper if you get to it, can you provide a sample of the |
please test |
The tests are being triggered in jenkins. |
+1 |
Comparison job queued. |
Comparison is ready Comparison Summary:
|
as suggested by @Sam-Harper , we are trying to test this PR on a Hilton machine (in my understanding, we would then be able to provide a sample of the |
@missirol thanks for the update! I'll try to pull some information out of the DQM online playback system in parallel. |
For the record: waiting for changes&validation at HLT. |
@missirol ran some tests but there were unrelated problems. Still waiting for HLT validation. |
The test by @missirol last week looks ok, so this is good to go. (Sorry, forgot to update this PR in time) |
+1 |
This pull request is fully signed and it will be integrated in one of the next master IBs (tests are also fine). This pull request will now be reviewed by the release team before it's merged. @davidlange6, @silviodonato, @fabiocos (and backports should be raised in the release meeting by the corresponding L2) |
+1 |
PR description:
In offline, the legacy TDirectory files for the DQMGUI are written by the
DQMFileSaver
module. While reading the code and its configurations, it became apparent that this module is not actually used in online or HLT (any more?), even though most of the code is dedicated to these cases. (Online and HLT useDQMServices/FileIO
now.)So, I stripped down this module to prove the hypothesis.
In addition, this PR contains the new TDirectory writing code by @andrius-k, as a new independent class. It is still linked into the DQMStore and can be called as
edm::Service<DQMStore>()->save(...)
. I also updated the Online output modules to use this helper (DQMFileSaverOnline
) or contain the writing code (DQMFileSaverPB
). Consequently,DQMStore
itself does not contain file writing code any more. There is still a huge amount of file reading code, which is largely useless and to be removed next.PR validation:
This PR needs testing in online DQM (not done yet) and ideally also in HLT (or a solid confirmation that the module is never loaded in HLT). The changes to the ProtoBuf writer need to be tested in the HLT/fastHadd/online DQM chain. None of this is done so far. I'll try to provide a test setup in another PR.
This PR needs testing in Tier0 Express processing, where the
runIsComplete
logic is used.The main point of the PR is to test
[I suspect some tests to fail (assuming these tests are run in the PR tests, not sure about that), and some flaws in the TDirectory writing code (it does not support QTests yet, for example).] Edit: all fixed.