Creation of Geometry Payloads for DataBase #46290
Comments
|
cms-bot internal usage |
|
A new Issue was created by @bsunanda. @Dr15Jones, @antoniovilela, @makortel, @mandrenguyen, @rappoccio, @sextonkennedy, @smuzaffar can you please review it and eventually sign/assign? Thanks. cms-bot commands are listed here |
|
assign geometry, alca, db |
|
New categories assigned: geometry,alca,db @atpathak,@bsunanda,@civanch,@consuegs,@Dr15Jones,@francescobrivio,@kpedro88,@makortel,@mdhildreth,@perrotta you have been requested to review this Pull request/Issue and eventually sign? Thanks |
cmsbuild
added
alca-pending
db-pending
geometry-pending
pending-signatures
and removed
pending-assignment
labels
|
assign core |
|
New categories assigned: core @Dr15Jones,@makortel,@smuzaffar you have been requested to review this Pull request/Issue and eventually sign? Thanks |
|
Thanks @perrotta for explaining the background and urgency. We'll try to take a look. It's unfortunate though the time line is so tight. |
|
I tried the instructions, and the first cmsRun job failed with and looking at $CMSSW_RELEASE/src/Geometry/CMSCommonData/data/dd4hep/ such a file doesn't exist. However, the following does cmsExtendedGeometry2024FlatPlus10Percent.xml |
|
Hi Chris
Did you start with a given IB of CMSSW? I wonder why FlatPlus10Percent appears 3 times in the name. I shall see what is there in the repository and make corrections. Best regards
Sunanda
…________________________________
From: Chris Jones ***@***.***>
Sent: 08 October 2024 18:24
To: cms-sw/cmssw ***@***.***>
Cc: Sunanda Banerjee ***@***.***>; Mention ***@***.***>
Subject: Re: [cms-sw/cmssw] Creation of Geometry Payloads for DataBase (Issue #46290)
I tried the instructions, and the first cmsRun job failed with
----- Begin Fatal Exception 08-Oct-2024 07:51:12 CDT-----------------------
An exception of category 'ConfigFileReadError' occurred while
[0] Processing the python configuration file named geometryExtended2024DD4hep_xmlwriter.py
Exception Message:
unknown python problem occurred.
RuntimeError: An exception of category 'FileInPathError' occurred.
Exception Message:
edm::FileInPath unable to find file Geometry/CMSCommonData/data/dd4hep/cmsExtendedGeometry2024FlatPlus10PercentFlatPlus10PercentFlatPlus10Percent.xml anywhere in the search path.
The search path is defined by: CMSSW_SEARCH_PATH
${CMSSW_SEARCH_PATH} is: [cut ]
and looking at $CMSSW_RELEASE/src/Geometry/CMSCommonData/data/dd4hep/ such a file doesn't exist. However, the following does cmsExtendedGeometry2024FlatPlus10Percent.xml
—
Reply to this email directly, view it on GitHub<#46290 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABGMZOWWDWJ3TFTZMVRWD3LZ2PIYXAVCNFSM6AAAAABPQEC2WOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGOJZG43DQNJZGY>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
|
Following @Dr15Jones suggestion in the private email thread, limiting VSIZE to ~5 GB ( |
|
Running again produces also an assertion failure i.e. here the fTopology.ncells() + fTopology.getHFSize() is 0.
This kind of varying behavior hints towards a memory corruption. |
|
So when I run the job (after fixing the scripts) I see the out of memory error. The debugger showed a very large allocation request. I then turned on one of the debug printouts and see Are these sizes what is actually expected? |
|
I got these numbers from the same printout (now that my test case went back to the A smoking gun in both the |
|
I understood that the script createExtended2024DD4hepPayloads.sh cannot be run multiple times in the same area because of sed. However, after running the script once cmsRun could be run multiple times and that is when I saw memory exhausted and got a system "kill". I am attaching a log file which I got for the first time. You could see "ERROR" getting printed multiple times starting from TKRECO_Geometry |
|
HCalTopology was not a new code. ZDCTopology is new and so is calowriters where the ZDC part is new. Maybe I remove the alignment part for ZDC and see the impact |
|
Got again the assertion failure behavior. This time the printout from This time the assertion failure stack trace was and problem being |
|
@makortel and I think we have found the problem. The values for cmssw/Geometry/CaloTopology/src/HcalTopology.cc Lines 67 to 83 in d8f3e55 Notice the if block. If neither of the two ifs are true, then HBSize_ is never set. Stepping through with the debugger shows that is the case here. I determined that the value of mode_ is 4 which corresponds to Run3 |
|
From git history I see the |
|
So after modifying the %MSG-s HCalGeom: HcalGeometryToDBEP:HcalGeometryToDBEP@callESModule 08-Oct-2024 09:51:58 CDT Run: 1 |
|
Thanks - I shall try to cure this |
|
I think the logic in HcalTopology needs to be modified |
|
See #46305 |
|
With the PR I made, the job still fails with |
|
Thanks Chris and Matti. There were some other logic in HcalTopology that were also wrong but the main issue was what you found. Thanks a lot
Sunanda
…________________________________
From: Chris Jones ***@***.***>
Sent: 08 October 2024 20:25
To: cms-sw/cmssw ***@***.***>
Cc: Sunanda Banerjee ***@***.***>; Mention ***@***.***>
Subject: Re: [cms-sw/cmssw] Creation of Geometry Payloads for DataBase (Issue #46290)
So after modifying the if block, I see what appears to be better values
%MSG-s HCalGeom: ***@***.*** 08-Oct-2024 09:51:58 CDT Run: 1
HcalGeometry_init(): HBSize 9216 HESize 14112 HOSize 2160 HFSize 7488
—
Reply to this email directly, view it on GitHub<#46290 (comment)>, or unsubscribe<https://github.com/notifications/unsubscribe-auth/ABGMZOSYE5MYSXBPNEZGDRTZ2PW7BAVCNFSM6AAAAABPQEC2WOVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDIMBQGA4DIOBWGU>.
You are receiving this because you were mentioned.Message ID: ***@***.***>
|
|
I was able to get the full script to run by adding to geometryExtended2024DD4hep_writer.py |
|
Thank you @makortel and @Dr15Jones for the big debug effort! |
|
Just for completeness, running valgrind (without #46305) did not reveal anything new. |
|
With the proposed corrections to HcalTopology (+ other changes needed to this class), and correcting the scenario description in Configuration/Geometry, the payload creation has been done for 2024. So this issue is resolved. |
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
The standard way of creating the payload is to follow these steps
cmsrel CMSSW_14_2_X_2024-10-06-2300
cd CMSSW_14_2_X_2024-10-06-2300/src
cmsenv
git cms-addpkg CondTools/Geometry
scram b -j4
cd CondTools/Geometry/test
/bin/cp writehelpers/* .
./createExtended2024DD4hepPayloads.sh 142DD4hepV1
This creates several .db files, some for XML files used for simulation and a number of files needed for loading parameters for reconstruction geometry
There are several cmsRun steps in createExtended2024DD4hepPayloads.sh
The second cmsRun which utilises cmsRun geometryExtended2024DD4heo_writer.py does not complete and gets killed.
Consequently, several .db files are not created which are recommended geometries for HCAL, ZDC, ,,,, and some parameters for Tracker
The text was updated successfully, but these errors were encountered: