Add FT0 module#536
Conversation
Barthelemy
left a comment
Barthelemy
left a comment
There was a problem hiding this comment.
Hi,
thank you.
I have a question: why not fill mChargeHistogram directly in the task ?
it seems like an extra step to first populate a TTree with every single event you get and then process the tree to populate the histogram.
Moreover, it seems that there mChargeHistogram in both the Task and the PostProcessing.
Otherwise I don't really have comments on the code itself.
| /// | ||
| /// \file ChannelsCheck.h | ||
| /// \author Milosz Filus | ||
| /// Dummy check for FT0 detector |
There was a problem hiding this comment.
Is it really a dummy check ?
| /// | ||
| /// \file ChannelsCheck.h | ||
| /// \author Milosz Filus | ||
| /// Another dummy check for FT0 detector |
|
Hello Barth, |
| @@ -0,0 +1,106 @@ | |||
| { | |||
There was a problem hiding this comment.
Hi, I saw on todays slides by Alla that you had a multinode workflow where you merge sampled data, instead of monitor objects. Is that really true, or that was some kind of mistake in slides?
There was a problem hiding this comment.
To be honest I'm a bit confused, because from generated graph https://i.imgur.com/sCPfHVe.jpg (using -g option) there is no task execution on local node (which would create monitoring object), but merger works on monitoring objects, so input of merger should be a monitoring object. Well I need to look on implementation because it looks like I'm missing something.
There was a problem hiding this comment.
Let us know if you need some clarifications.
There was a problem hiding this comment.
See the slide 5, this is how it is supposed to look like in the big picture:
https://indico.cern.ch/event/873380/contributions/3850119/attachments/2036439/3409783/WP7plenaryMay.pdf
Some parallel proxies are replaced with one, as far as I remember, but otherwise it should match what you get (disregarding devices generated automatically by DPL).
There was a problem hiding this comment.
@knopers8
Thank you for the slide with explanation.
Right, it seems I made a mistake in a slide, there is a task execution and monitoring object creation before sending it to remote device, but I do not undestand why the generated graph does not contain task runner.
Maybe adding -g option ends workflow generation before attaching task runner to workflow?
There was a problem hiding this comment.
I doubt it. I've just crosschecked on my machine and the proxy after dispatcher is still created.
In case of remote tasks: Does your machine name (the --host argument, specifically) match one from your list of machines in a data sampling policy? A proxy for sampling policies won't be created on machines which don't need it.
Similarly, in case of local tasks, you need your machine name among "machines" in the task configuration.
There was a problem hiding this comment.
@knopers8 Well I will update multinode-config to my current version, because I completely removed data sampling policies + remote task.
Sorry, I was describing my current version, forgot to update this PR.
There was a problem hiding this comment.
Yes, so to make your local task BasicDigitQcTask appear, you have to run QC with --local --host 2a (or 3a), so QC knows that this is the proper machine to load this task. I am not sure however, how far can you get away with mocking the names instead of using real machines.
3 participants
As requested, first part of QC for FT0 detector.
It is worth to mention that there is a still unsolved issue which may affect this code.
https://alice.its.cern.ch/jira/browse/QC-459