[OPNFLWPLUG-850] [Boron] Openflow plugin deletes switch from topology while changing mastership from one controller to another Created: 09/Feb/17  Updated: 27/Sep/21  Resolved: 18/Sep/17

Status: Resolved
Project: OpenFlowPlugin
Component/s: General
Affects Version/s: None
Fix Version/s: None

Type: Bug
Reporter: Jon Castro Assignee: Luis Gomez
Resolution: Done Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified
Environment:

Operating System: All
Platform: All


External issue ID: 7763

 Description   

A controller that holds mastership to a switch and lose connectivity delete the switch from the topology even though there are other controllers which will valid connectivity and another controller takes ownership.

Reproducing this error is very straightforward:

  • Start a 3 controllers nodes
  • connect a switch to all controllers
  • delete the controller that holds mastership from the switch

Then, switch disappears from the topology and appears again when another controllers takes ownership.



 Comments   
Comment by Luis Gomez [ 09/Feb/17 ]

I think this issue will be adressed here:

https://git.opendaylight.org/gerrit/#/c/51071/

BR/Luis

Comment by Luis Gomez [ 09/Feb/17 ]

To be more precise, we need all these patches merged:
https://git.opendaylight.org/gerrit/#/q/topic:%22Bug+6802%22

Comment by Anil Vishnoi [ 27/Feb/17 ]

stable/boron : https://git.opendaylight.org/gerrit/#/c/52230/

Comment by Jozef Bacigal [ 01/Mar/17 ]

In Carbon release I got promise from mdsal guys to change the API on singleton service, so we don't need to combine EOS and singleton. (Plan is RC0)

Comment by Anil Vishnoi [ 02/Mar/17 ]

RC0 probably is going to be too late for us. We can use this fix till we get the singleton clustering api changes. We can take a call at that time.

Comment by Abhijit Kumbhare [ 23/Mar/17 ]

This is needed fixed in master.

Comment by Jozef Bacigal [ 25/Apr/17 ]

Luis, is this still an issue ?

Which test should I look at ?

Comment by Luis Gomez [ 25/Apr/17 ]

There is no system test for this, I think it is better to ask Jon directly.

Comment by Jon Castro [ 25/Apr/17 ]

Is this bug solved on Master branch? I could tested if the fix is merched to Boron branch.

Comment by Luis Gomez [ 25/Apr/17 ]

Looking at the bug history, it seems Boron is the fixed branch and there is some promise to fix for Carbon too.

Comment by Jozef Bacigal [ 27/Apr/17 ]

Jon, I will let you know when I cherry-pick patches from carbon to boron.

Comment by Jon Castro [ 28/Apr/17 ]

I have done tests with latest Boron (SR3) but switches does not move from one controller to another without topology/flows changes.

I detected that master branch contains some commits that may be related to this bug but does not seem to be merged in carbon and Boron

https://git.opendaylight.org/gerrit/#/q/topic:%22Split+connection+layer+...%22

Comment by Luis Gomez [ 28/Apr/17 ]

Anil, is your patch: https://git.opendaylight.org/gerrit/#/c/52230 doing something for this issue or is it really the split connection patches in master that help here?

Comment by Anil Vishnoi [ 04/May/17 ]

@luis, Yes https://git.opendaylight.org/gerrit/#/c/52230/3 This patch should fix this behavior. I believe jozef is doing similar fix to the stable/carbon branch now.

Comment by Anil Vishnoi [ 04/May/17 ]

@luis, Yes https://git.opendaylight.org/gerrit/#/c/52230/3 This patch should fix this behavior. I believe jozef is doing similar fix to the stable/carbon branch now.

Comment by Abhijit Kumbhare [ 04/May/17 ]

(In reply to Jon Castro from comment #12)
> I have done tests with latest Boron (SR3) but switches does not move from
> one controller to another without topology/flows changes.
>
> I detected that master branch contains some commits that may be related to
> this bug but does not seem to be merged in carbon and Boron
>
> https://git.opendaylight.org/gerrit/#/q/topic:%22Split+connection+layer+...
> %22

These have been merged in master and in stable/carbon (but not yet in stable/boron). Can you please test it in stable/carbon so we can know whether the Split Connection Handler fixes this issue?

Comment by Luis Gomez [ 15/May/17 ]

The question is for Jon again as we do not have system test for this.

Comment by Luis Gomez [ 15/May/17 ]

@Anil, https://git.opendaylight.org/gerrit/#/c/52230/3 was merged back in March well before SR3 but Jon said above Boron SR3 does not work, so there is some confusion here whether the issue is addressed in your patch or in the split connection patches.

Comment by Anil Vishnoi [ 15/May/17 ]

(In reply to Luis Gomez from comment #18)
> @Anil, https://git.opendaylight.org/gerrit/#/c/52230/3 was merged back in
> March well before SR3 but Jon said above Boron SR3 does not work, so there
> is some confusion here whether the issue is addressed in your patch or in
> the split connection patches.

In stable/boron, above patch supposed to fix this issue. In stable/carbon, split connection patches contains the fix for this issue.

I will test with stable/boron and confirm if this issue is resolved or not.

Comment by Luis Gomez [ 01/Aug/17 ]

I think this should be fixed now, please double check.

Comment by Abhijit Kumbhare [ 18/Sep/17 ]

This has been fixed for a while

Generated at Wed Feb 07 20:33:32 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.