Comment 1 for bug 1747708

It looks like when the charm calls status-set, the 'since' field in the application and unit status is updated, even if there was no change made.

Here are two batches of the notifications I'm seeing sent every five minutes from a minimal deployment. The deltas are identical, apart from the since field in the statuses:

[application change map[subordinate:false charm-url:cs:postgresql-165 owner-tag: min-units:0 life:alive constraints:map[] status:map[current:active message:Live master (9.5.10) since:2018-02-06T16:34:06.401035688Z version:] workload-version:9.5.10 model-uuid:94f4722f-4a1e-47d0-8fe8-985f9f45da08 name:postgresql exposed:false]]
[unit change map[series:xenial charm-url:cs:postgresql-165 private-address:10.0.4.244 machine-id:0 subordinate:false workload-status:map[version: current:active message:Live master (9.5.10) since:2018-02-06T16:34:06.401035688Z] name:postgresql/0 application:postgresql public-address:10.0.4.244 ports:[map[protocol:tcp number:5432]] port-ranges:[map[from-port:5432 to-port:5432 protocol:tcp]] agent-status:map[current:idle message: since:2018-02-06T16:29:53.494424567Z version:] model-uuid:94f4722f-4a1e-47d0-8fe8-985f9f45da08]]

[application change map[model-uuid:94f4722f-4a1e-47d0-8fe8-985f9f45da08 exposed:false owner-tag: status:map[version: current:active message:Live master (9.5.10) since:2018-02-06T16:38:33.795944943Z] workload-version:9.5.10 subordinate:false name:postgresql charm-url:cs:postgresql-165 life:alive min-units:0 constraints:map[]]]
[unit change map[application:postgresql series:xenial public-address:10.0.4.244 private-address:10.0.4.244 ports:[map[protocol:tcp number:5432]] subordinate:false workload-status:map[current:active message:Live master (9.5.10) since:2018-02-06T16:38:33.795944943Z version:] agent-status:map[message: since:2018-02-06T16:29:53.494424567Z version: current:idle] model-uuid:94f4722f-4a1e-47d0-8fe8-985f9f45da08 name:postgresql/0 charm-url:cs:postgresql-165 machine-id:0 port-ranges:[map[protocol:tcp from-port:5432 to-port:5432]]]]

So this may be a bug in status handling, rather than the WatchAllModels call. It is very common for charms to blindly reset their workload status.