Added support for 'grid_accounting' daemon

[Martin-Rehr]

The 'grid_accounting' daemon generate user storage accounting based on 'home', 'vgrid' and 'peers' usage.

[jonasbardino]

Generally looks good. I've commented in-line for now.

[jonasbardino]

I just realized I hadn't seen mig/lib/accounting.py at all during my review because github thinks it's too big to display the diff. At a first glance I didn't see any obvious issues with it but perhaps just a couple of questions or remarks that you may or may not already have covered.

1. Users cannot in practice be both external and act as peers. The comments made me unsure if that is already taken care of. Any external users with leftover peers settings (from a past as internals or whatever) could probably just be ignored in the peers mapping.
2. Multiple internal users can in practice accept the same external user as peer. I guess only one should then be accounted for the disk use of that external user, but so far didn't find where the accounting is then in fact assigned to only the 'active' peer or if it perhaps silently gets truncated to the last one found in the peers mapping.

We can proceed with the merge and address any pending corner-cases as follow-up if you prefer.

[Martin-Rehr]

Thanks for the review:

1. In practice we have users that occur as both external and act as peers. A log warning is thrown in that case:
   

   migrid-sync/mig/lib/accounting.py

   Line 659 in 08d6799

   if ext_users and peers:

Which gives us the information needed for a manual cleanup

2. In this initial version all internal users that accepted the external user X as peer are accounted for the external user X

[jonasbardino]

Thanks for the review:

1. In practice we have users that occur as both external and act as peers.  A log warning is thrown in that case:
   https://github.com/ucphhpc/migrid-sync/blob/08d6799b4345c5a9f6953537ce8c405746a12336/mig/lib/accounting.py#L659

Which gives us the information needed for a manual cleanup

2. In this initial version all internal users that accepted the external user X as peer are accounted for the external user X

Thanks for the clarification 👍

[jonasbardino]

Not sure if you just need to resolve the open comment about unit tests or if you have more updates in the pipeline but let's try to get it in when you're done.

[Martin-Rehr]

Not sure if you just need to resolve the open comment about unit tests or if you have more updates in the pipeline but let's try to get it in when you're done.

I have no further improvements for it in the pipeline, so If everybody can live with the latest version I'll mark it as resolved.