Ticket 1062

Summary:	Documentation update for HDF5 accounting
Product:	Slurm	Reporter:	Kilian Cavalotti <kilian>
Component:	Documentation	Assignee:	Danny Auble <da>
Status:	RESOLVED FIXED	QA Contact:
Severity:	5 - Enhancement
Priority:	---	CC:	da
Version:	14.03.6
Hardware:	Linux
OS:	Linux
Site:	Stanford	Alineos Sites:	---
Atos/Eviden Sites:	---	Confidential Site:	---
Coreweave sites:	---	Cray Sites:	---
DS9 clusters:	---	HPCnow Sites:	---
HPE Sites:	---	IBM Sites:	---
NOAA SIte:	---	OCF Sites:	---
Recursion Pharma Sites:	---	SFW Sites:	---
SNIC sites:	---	Linux Distro:	---
Machine Name:		CLE Version:
Version Fixed:	14.03.8 14.11.0pre5	Target Release:	---
DevPrio:	---	Emory-Cloud Sites:	---
Attachments:	HDF5 accounting doc update HDF5 accounting doc update

Description Kilian Cavalotti 2014-08-22 09:44:18 MDT

Hi, 

Trying the HDF5 accounting capabilities, and I think a couple of updates are needed in the docs. Could you please check if I'm correct?

Thanks.

Comment 1 Kilian Cavalotti 2014-08-22 09:45:02 MDT

Created attachment 1157 [details]
HDF5 accounting doc update

Comment 2 Kilian Cavalotti 2014-08-22 09:51:03 MDT

Created attachment 1158 [details]
HDF5 accounting doc update

I missed a couple things in the first one

Comment 3 Danny Auble 2014-08-22 10:18:13 MDT

Thanks Kilian, some of these are correct, but not all.

I believe JobAcctGatherFrequency is correct in the hdf5 user guide, please see the slurm.conf man page for AcctGatherNodeFreq and JobAcctGatherFrequency.  If I am wrong please let me know what issues you are seeing with it.

The Network => Lustre changes and the Luster => Lustre mistakes are correct.  You caught a couple of cluster -> clustre changes though ;).

So my watered down patches are in commits

4768bd96cc0f73810169a53b766eb1a06fb012a7
ba7f5cb20957927601455484b3b11c5bab73c4a4

Let me know if I missed anything

Comment 4 Kilian Cavalotti 2014-08-22 10:27:48 MDT

Hi Danny,

(In reply to Danny Auble from comment #3)
> I believe JobAcctGatherFrequency is correct in the hdf5 user guide, please
> see the slurm.conf man page for AcctGatherNodeFreq and
> JobAcctGatherFrequency.  If I am wrong please let me know what issues you
> are seeing with it.

Oh ok, I was under the impression that JobAcctGatherFrequency was only controlling jobs accounting used by the JobAcctGatherType plugin, and that the other accounting plugins (AcctGatherEnergyType, AcctGatherInfinibandType and AcctGatherFilesystemType) were controlled by AcctGatherNodeFreq. That's why I thought the hdf5 user guide should mention AcctGatherNodeFreq instead of JobAcctGatherFrequency. But I didn't do any testing to support this. 

> The Network => Lustre changes and the Luster => Lustre mistakes are correct.
> You caught a couple of cluster -> clustre changes though ;).

Damn, that was a sloppy job, sorry...
 
> So my watered down patches are in commits
> 
> 4768bd96cc0f73810169a53b766eb1a06fb012a7
> ba7f5cb20957927601455484b3b11c5bab73c4a4
> 
> Let me know if I missed anything

Looks great, thank you!