Still while investigating my broken builds, I see that the builds running on Travis are unreliable.
I believe that this is related to the amount of processing needed to start the containers completely. I see that this has probably already been an issue in the past when I see all the "sleep" commands present in the Makefile and in some tests.
The following screen show illustrates it at best:
Source: https://travis-ci.org/callmemagnus/docker-mailserver/branches
Both branches are the same. "Patch-1" is the source of my PR and its build did not trigger on your repo (?).
I am pretty convinced that the result of the builds are dependent on the availability and power of Travis build slave.
Still while investigating my broken builds, I see that the builds running on Travis are unreliable.
I believe that this is related to the amount of processing needed to start the containers completely. I see that this has probably already been an issue in the past when I see all the "sleep" commands present in the Makefile and in some tests.
The following screen show illustrates it at best:
Source: https://travis-ci.org/callmemagnus/docker-mailserver/branches
Both branches are the same. "Patch-1" is the source of my PR and its build did not trigger on your repo (?).
I am pretty convinced that the result of the builds are dependent on the availability and power of Travis build slave.