Ticket #345 (closed defect: fixed)

Opened 5 years ago

Last modified 5 years ago

Unable to contact SM

Reported by: ibaldin Owned by: ibaldin
Priority: major Milestone:
Component: ORCA: controllers - ORCA API Version: baseline
Keywords: Cc: yxin, anirban, pruth

Description

From Tahsin:

Today, I am constantly getting the following exception when I tried to create a slice across two racks from a saved request file. It was not happening yesterday with the same file. I then manually created an topology across two racks and it worked. Just letting you know since it could be an issue. The topology is attached with the email and it had Node 3 and Node 7 in Duke CS rack while Node 2 and Node 4 on BBN rack.

I also faced the same error a while ago with simple hand drawn topology using two racks on flukes.

(images and RDF files attached)

Attachments

slice-1.png (33.2 kB) - added by ibaldin 5 years ago.
Slice picture
error.png (20.4 kB) - added by ibaldin 5 years ago.
Error message
4 ring, 2 diff racks (15.6 kB) - added by ibaldin 5 years ago.
Saved RDF file

Change History

Changed 5 years ago by ibaldin

Slice picture

Changed 5 years ago by ibaldin

Error message

Changed 5 years ago by ibaldin

Saved RDF file

Changed 5 years ago by ibaldin

  • summary changed from Likely flukes save/restore issue to Unable to contact SM

This has been reported by Yufeng in some cases on geni-test. Still not clear what is going on. This is not Flukes, this is controller not being able to talk to SM.

Changed 5 years ago by ibaldin

  • cc yxin, anirban, pruth added

Changed 5 years ago by ibaldin

I found that in a bunch of places where getSM() was called on XmlrpcOrcaState?, there was a missing call to returnSM(sm). This is necessary, because getSM gets a thread from a pool, which must be returned. It is possible that these errors had to do with SM thread not being returned into the pool. I have no gone through the code and added a matching returnSM call to every getSM call.

Changed 5 years ago by ibaldin

On r6677

Changed 5 years ago by ibaldin

Have we seen this recently? In emulation or for real?

Changed 5 years ago by ibaldin

  • component changed from External: FLUKES to ORCA: controllers - ORCA API

Changed 5 years ago by ibaldin

  • status changed from new to closed
  • resolution set to fixed

Closing - appears fixed.

Note: See TracTickets for help on using tickets.