When specifying a requisite and you only need to list a single item you
can specify it without being wraped in a list and we will handle it
automatically. This was being handled when we built our attrs to output
the final data, but this failed when you extended a require requisite
inside of a context manager.
From the Python docs on the exec statement:
> Remember that at module level, globals and locals are the same dictionary.
> If two separate objects are given as globals and locals, the code will be
> executed as if it were embedded in a class definition.
We were providing a specific object for locals and in the specific case
reported in #21796 this caused a very strange name error when used in a
specific way. By removing the explicit locals dictionary and just having the
globals dictionary be shared fixes the issue, and we weren't using the
specific locals anyway.
From the Python docs on the exec statement:
> Remember that at module level, globals and locals are the same dictionary.
> If two separate objects are given as globals and locals, the code will be
> executed as if it were embedded in a class definition.
We were providing a specific object for locals and in the specific case
reported in #21796 this caused a very strange name error when used in a
specific way. By removing the explicit locals dictionary and just having the
globals dictionary be shared fixes the issue, and we weren't using the
specific locals anyway.
Fixes#21707
The issue is basically that the master has seen N minions-- and not all N are currently connected. The publish job returns to the salt-api that N minions got the job-- and it dutifilly waits for all the returns. With this we use the same ping timeout that the CLI does while waiting for returns. So once all minions are no longer running the job we will return.
Fixes#21707
The issue is basically that the master has seen N minions-- and not all N are currently connected. The publish job returns to the salt-api that N minions got the job-- and it dutifilly waits for all the returns. With this we use the same ping timeout that the CLI does while waiting for returns. So once all minions are no longer running the job we will return.
