The logs we have (logsink from machine 2) have entries indicating that some units in a cross model relation are entering the scope of the relation, which is a prerequisite for running the relation-joined hook.
eg
2021-08-26 02:18:59 DEBUG juju.apiserver.common.crossmodel crossmodel.go:124 remote application for changed relation remote-5816942fe3a54d858170eb1f7fd25eb9:ceph use01-ceph-mon:client is remote-5816942fe3a54d858170eb1f7fd25eb9 in model b5c17291-b581-4144-8148-a81c1e9cd4dd
2021-08-26 02:18:59 DEBUG juju.apiserver.common.crossmodel crossmodel.go:166 remote-5816942fe3a54d858170eb1f7fd25eb9/18 entering scope (map[egress-subnets:172.25.80.132/32 ingress-address:172.25.80.132 private-address:172.25.80.132])
Note
"..... remote application for changed relation xxxx ..."
and
" xxxx/0 entering scope ...."
The thing is that these are for the offering model. I don't see any corresponding log entries to say that the (say ceph related) units on the consuming side have received change events. It could be that one or more the issues with "not found" errors is preventing the event processing from completing, though am not 100% sure yet.
The logs we have (logsink from machine 2) have entries indicating that some units in a cross model relation are entering the scope of the relation, which is a prerequisite for running the relation-joined hook.
eg
2021-08-26 02:18:59 DEBUG juju.apiserver. common. crossmodel crossmodel.go:124 remote application for changed relation remote- 5816942fe3a54d8 58170eb1f7fd25e b9:ceph use01-ceph- mon:client is remote- 5816942fe3a54d8 58170eb1f7fd25e b9 in model b5c17291- b581-4144- 8148-a81c1e9cd4 dd common. crossmodel crossmodel.go:166 remote- 5816942fe3a54d8 58170eb1f7fd25e b9/18 entering scope (map[egress- subnets: 172.25. 80.132/ 32 ingress- address: 172.25. 80.132 private- address: 172.25. 80.132] )
2021-08-26 02:18:59 DEBUG juju.apiserver.
Note
"..... remote application for changed relation xxxx ..."
and
" xxxx/0 entering scope ...."
The thing is that these are for the offering model. I don't see any corresponding log entries to say that the (say ceph related) units on the consuming side have received change events. It could be that one or more the issues with "not found" errors is preventing the event processing from completing, though am not 100% sure yet.