weave-kube peer received conflicting IPAM update

[bboreham]

Reported by user "fly" on Slack

Symptom was that the same IP address had been allocated on two pods.

Attached files are weave report from both nodes and logs from node2. We do not have the log from node1.

We can see in the report for node1 it has a conflicting view of the IP allocation data and is rejecting connections from other Weave peers:

"Info": "Received update for IP range I own at 172.29.16.0 v31: incoming message says owner 0a:a9:1b:94:f0:e7 v35

Somehow node1 is at version 31 while other nodes have v35.

User had done two restarts, for a Weave version upgrade and to turn on encryption. Both of these would likely delete the persistence file, which is #2610.

If the persistence file was deleted, the conflicting versions must have come from other peers. Perhaps the partition created by adding encryption creates this situation?

weave-net-ms22r_kube-system_weave-6ec2590fec39203f88d200a60e77bd22a36eba38bd41552aaf13fe35efb08f2d.log.gz
weave-node-reports.zip

[bboreham]

I'm going to close this as a duplicate of #2610; we have had no further reports or other information that would indicate otherwise