Add CTest for XRootd Remote support

[eisenhauer]

No description provided.

[vicentebolea]

@eisenhauer I added tmp images to check that xrootd is installed correctly

[eisenhauer]

Cool. Lets see how this goes...

[vicentebolea]

@eisenhauer find the temp images with xrootd.

[eisenhauer]

@vicentebolea Hey Vicente! So current issue is that the XRootD server really doesn't want to run as root. It can be started as root and you can use the -R option to tell it another user or UID to run as, but "-R root" or "-R 0" are rejected. Are there github action options that we might use to help get around this? I.E. adding a user to the docker that we only use to run the Xrootd server?

[eisenhauer]

Digging into the XRootD code here, this error means that getpwnam("daemon") didn't return anything... If you pass a numeric value it passes that to getpwuid().

[vicentebolea]

@eisenhauer the tests finally passes

[eisenhauer]

Nice!

[eisenhauer]

@vicentebolea Time to revive this PR and prepare to merge. I tried (badly) to update the CI to reflect changes since it was first done, but it likely needs you attention.

[vicentebolea]

do you need to re-do the docker images?

[eisenhauer]

I'm actually not sure what you did before to get these to run. Temporary docker images? There were conflicts with master in some of the setup files that I tried to resolve, but perhaps incorrectly.

[eisenhauer]

So I guess the goal is this: The non-CI stuff in here mostly just enables testing of the XRootD plugin. We'd like to test on one or more platform in CI, so we'd need xrootd installed. Doesn't have to be everywhere, but one or two places would be good.

[vicentebolea]

@eisenhauer the CI is ready, the 20.04 images now have xrootd.

[eisenhauer]

Minor tweak to fix server args.

[vicentebolea]

Added a few comments, looks good otherwise. After addressing the comemnts that makes sense to you I can remove the 'tmp' and we can merge this.

[vicentebolea]

@eisenhauer we are still having errors, let me know when the CI is green

[eisenhauer]

@vicentebolea It looks like we're back to the user problem. In the manipulations, did we lose the bit that made this work?

[vicentebolea]

@eisenhauer lets squash-merge to get rid of all this fixup commits

[vicentebolea]

@eisenhauer images are pushed (without the tmp)