Error running migrations in CI

[haikyuu]

Here is the error I get in this workflow when running migrations

....

2021-09-07 14:20:59.749 UTC [3999] NOTICE:  function edgedbpub.bXjme94YjAMRx9AIZwtOnA:ab150ea_target-del-def-inl-f() does not exist, skipping
2021-09-07 14:20:59.749 UTC [3999] CONTEXT:  SQL statement "DROP FUNCTION IF EXISTS edgedbpub."bXjme94YjAMRx9AIZwtOnA:ab150ea_target-del-def-inl-f"()"
	PL/pgSQL function inline_code_block line 1040 at SQL statement
Applied m17n33vyk6esc7ay4i2h76rw42llkuewlemjnlfbn75cf5hiklmnga (00002.edgeql)
Applied m1iepbg6q65oubedirjc226akcu57tspdn23x6pdlxclraac6knila (00003.edgeql)
Applied m1ty4xwamqqhasln23rczlwrursrwkhhrspanryltu4adq6yzxvsbq (00004.edgeql)
Applied m1xunomwswwrf7iskdxsekf4uft6jpeapuzhkplajty4qdjq5uk2fq (00005.edgeql)
Applied m1oz2tq5hxskt4csdyqjxjs76txp2jenpurijmge2kh67eqbkb56xq (00006.edgeql)
INFO 3977 2021-09-07 14:21:01,901 edb.server: Received signal: Signals.SIGTERM.
INFO 3977 2021-09-07 14:21:01,901 edb.server: Shutting down.
2021-09-07 14:21:01.933 UTC [3986] LOG:  received fast shutdown request
2021-09-07 14:21:01.936 UTC [3986] LOG:  aborting any active transactions
2021-09-07 14:21:01.937 UTC [3986] LOG:  background worker "logical replication launcher" (PID 3993) exited with exit code 1
2021-09-07 14:21:01.937 UTC [3999] FATAL:  terminating connection due to administrator command
2021-09-07 14:21:01.937 UTC [3999] LOG:  could not send data to client: Broken pipe
2021-09-07 14:21:01.944 UTC [3988] LOG:  shutting down
2021-09-07 14:21:01.969 UTC [3986] LOG:  database system is shut down
Bootstrapping complete, but there was an error creating the service. You can start it manually via: 
  edgedb instance start --foreground perfect_stack

[1st1]

@fantix can you take a look at this?

[fantix]

Looks like the issue is hidden in earlier logs:

Failed to connect to bus: No such file or directory
...
edgedb error: cannot create service: process "systemctl" "--user" "daemon-reload" failed: exit status: 1

In short, this is an issue that the EdgeDB CLI tries to use systemd for managing instances but failed to do so.

Someone succeeded with some workarounds, while there is an open PR edgedb/setup-edgedb#22 which - when done - may work for this scenario.

[haikyuu]

I'll try again when that PR is merged @fantix