[CONTROLLER-1843] Spurious mdsal events after leader taken down Created: 25/Jun/18  Updated: 25/Jun/18  Resolved: 25/Jun/18

Status: Resolved
Project: controller
Component/s: None
Affects Version/s: None
Fix Version/s: None

Type: Bug Priority: Medium
Reporter: Josh Hershberg Assignee: Josh Hershberg
Resolution: Won't Do Votes: 0
Labels: None
Remaining Estimate: Not Specified
Time Spent: Not Specified
Original Estimate: Not Specified


 Description   

This issue was observed in the process of netvirt development. When the configuration leader is taken down objects in the datastore fire DTCL events even though they were created long ago.

Some more details on this here: https://bugzilla.redhat.com/show_bug.cgi?id=1567128#c9

 

 

 Comments   
Comment by Tom Pantelis [ 25/Jun/18 ]

I assume the DTCNs were fired on a different node which took over as leader. And I assume the DTCL is not a ClusteredDataTreeChangeListener in which case the DTCL is only notified on the leader. So when a new leader takes over, the DTCL is then registered which notifies of all initial data. Sounds like that's what you're seeing - if so, it's working as designed.

Comment by Josh Hershberg [ 25/Jun/18 ]

Turns out this is not a bug at all. See Tom's comment

Comment by Tom Pantelis [ 25/Jun/18 ]

Well that depends on how you look at it  This part was always sticky and was implemented before the ClusteredDTCL was conceived. Since non-clustered DTCL ensures only one notification occurs in the clustered,  one could argue it should do the same for the initial pre-existing data update.

Generated at Wed Feb 07 19:56:35 UTC 2024 using Jira 8.20.10#820010-sha1:ace47f9899e9ee25d7157d59aa17ab06aee30d3d.