Opened 12 years ago

Closed 11 years ago

Last modified 11 years ago

#115 closed (fixed)

Requesting a resource that is being allocation to some other sliver causes other sliver to fail

Reported by: Owned by: somebody
Priority: major Milestone:
Component: AM Version: SPIRAL4
Keywords: Cc:


Ran a test case which requested the 1 bare metal resource while it was being allocated to a separate sliver. This request caused for the other existing sliver request to fail.

Here is an outline for this failure:

  1. Created 1st sliver with 1 bare metal node and 1 VM.
  2. While the 1st sliver was being created (about 1 min after step 1) created a 2nd slice.
  3. Created a 2nd sliver requested for the same bare metal node and 1 VM.


The 2nd sliver request never failed, it quickly reported success and a sliver manifest. The sliver status showed the bare metal as "Ticketed" and remained in this state.

The 1st sliver creation continued to run for about 5 more minutes and then failed with the following error:

Result Summary: Failed CreateSliver for slice errors at  Error from Aggregate: code 2: Request id: null
Embedding workflow ERROR: 1:Insufficient resources or Unknown domain:! Requested: baremetalce:1. 

Shouldn't the bare metal be unavailable, while the 1st sliver is being created?

Change History (5)

comment:1 Changed 12 years ago by

As the user that has created the 1st sliver which failed, requesting a sliver status after the failure reports:

Failed to get SliverStatus on errors at AM Error from Aggregate: code 2: ERROR: unable to get slice status for

The sliver should not exist after the failure, right? Expected the sliverstatus to

comment:2 Changed 11 years ago by


Could you please re-test and see if this remains valid?

comment:3 Changed 11 years ago by

Sure trying it now.

"Luisa" :-)

comment:4 Changed 11 years ago by

Resolution: fixed
Status: newclosed

Tried several times to request a bare metal node while it was being provisioned to another slice, and each time got back an appropriate failure "Insufficient resources for specified". Also each conflicting request did not impact the sliver creation that was already in progress.

This problem is solved, closing ticket.

comment:5 Changed 11 years ago by

Apologies for the misspelling. ;)

Note: See TracTickets for help on using tickets.