Fail to update watchtower when http api is running

[yrien30]

Describe the bug

Watchtower does not seem to be able to create a new instance of watchtower because http api port is allready used by current watchtower instance.

time="2021-05-04T06:41:49Z" level=debug msg="Renaming container /watchtower (7466244452a6) to bVxoHrZQQfmJwmVPzqUzmrEWtaigXveU"                                                                                                                              │
│   time="2021-05-04T06:41:49Z" level=info msg="Creating /watchtower"                                                                                                                                                                                            │
│   time="2021-05-04T06:41:51Z" level=debug msg="Starting container /watchtower (f2211634e4c3)"                                                                                                                                                                  │
│   time="2021-05-04T06:41:52Z" level=error msg="Error response from daemon: driver failed programming external connectivity on endpoint watchtower (0c44b36fce32cb7f8b15a6ae631577478607525311c1dcd03adebe017933944e): Bind for 0.0.0.0:8080 failed: port is    │
│   already allocated" 

To Reproduce
Enable http api:
-p 8080:8080 -e WATCHTOWER_HTTP_API_UPDATE=true -e WATCHTOWER_HTTP_API_TOKEN=Token

[simskij]

Interesting! Thank you for the report! I'll add it to the 1.4 milestone

[piksel]

Highly related to #939
I added an endpoint that just closed down the API (given the right token of course) and that just halted the container (since the only thing blocking it is the http server).
Update: And even more confusing, if the http.ErrorServerClosed error is ignored (which it should be), it starts the scheduler upon closing the web API.