Ticket 433

Summary:	Incorrect depth for AcceleratorAllocation in parser_basil_5.1.c
Product:	Slurm	Reporter:	David Gloe <david.gloe>
Component:	Cray ALPS	Assignee:	Moe Jette <jette>
Status:	RESOLVED DUPLICATE	QA Contact:
Severity:	2 - High Impact
Priority:	---	CC:	da
Version:	2.6.x
Hardware:	Linux
OS:	Linux
Site:	CRAY	Slinky Site:	---
Alineos Sites:	---	Atos/Eviden Sites:	---
Confidential Site:	---	Coreweave sites:	---
Cray Sites:	---	DS9 clusters:	---
Google sites:	---	HPCnow Sites:	---
HPE Sites:	---	IBM Sites:	---
NOAA SIte:	---	NoveTech Sites:	---
Nvidia HWinf-CS Sites:	---	OCF Sites:	---
Recursion Pharma Sites:	---	SFW Sites:	---
SNIC sites:	---	Tzag Elita Sites:	---
Linux Distro:	---	Machine Name:
CLE Version:		Version Fixed:
Target Release:	---	DevPrio:	---
Emory-Cloud Sites:	---
Attachments:	Patch which should fix the issue

Description David Gloe 2013-09-30 06:17:49 MDT

Ian Ryan at Cray reported this issue:
"I was trying to build/install a hybrid alps/slurm, with the new slurm 2.6.2.
I can get one job to run but the server crashes after the job returns. This is the message in the server log:

[2013-09-30T11:39:49.003] fatal: Tag 'AcceleratorAllocation' appeared at depth 7 instead of 5

If I search for 'AcceleratorAllocation' in the source I find it in an alps plugin file. Any idea why I’m getting this error?"

It seems that in parser_basil_5.1.c, AcceleratorAllocation is listed at depth 5 rather than depth 7. Depth 7 is correct, since it should appear at 
ResponseData->Inventory->NodeArray->Node->AcceleratorArray->Accelerator->AcceleratorAllocation by the BASIL 1.2 specification.

The depth is correct in parser_basil_4.0.c. I've attached a patch which should fix this issue.

Comment 1 David Gloe 2013-09-30 06:18:54 MDT

Created attachment 420 [details]
Patch which should fix the issue

Comment 2 Danny Auble 2013-09-30 06:23:31 MDT

It would appear I guessed wrong when dealing with this in the orignal bug on the matter, I'll change it back.

Thanks for reporting

*** This ticket has been marked as a duplicate of ticket 395 ***

Comment 3 Danny Auble 2013-09-30 06:26:36 MDT

David, could you send me the document you refer to here on the spec of BASIL 1.2.  I feel things changed in 1.3.  If you happen to have the spec file fro BASIL 1.3 that would be helpful on the topic.

Comment 4 Danny Auble 2013-09-30 06:30:32 MDT

What version of CLE is this person running.  You reference BASIL 1.2 but anything running a CLE version of 5.1 should allow BASIL 1.3 to work and that is what we are using.

Comment 5 David Gloe 2013-09-30 06:34:22 MDT

As far as I know the location of the AcceleratorAllocation hasn't changed between BASIL 1.2 and 1.3. I mentioned 1.2 because that's where the AcceleratorAllocation element was added; it is still supported in BASIL 1.3, and as far as I know the depth hasn't changed.

It looks like your original patch moved AcceleratorArray to 5 and Accelerator to 6, but neglected to move AcceleratorAllocation to a child of Accelerator at 7.

I'll send the 1.2 and 1.3 documentation I have to you in an email.

Comment 6 Danny Auble 2013-09-30 06:40:59 MDT

The original patch moves the AcceleratorAllocation from 7 to 5 but it was never tested.  I just guessed it was that way since the other Accelerator depths changed.

I am moving future comments to 395 so we can just deal with one bug.