>> In my understanding MirrorBrain is lacking proper job queue and adding
>> it + WebUI would result in managing two different projects inside
>> single github repository. This is why we have two projects. I would be
>> glad to do full help if somebody had a good vision how to fix it in
>> mirrorbrain.

> Thinking out loud - the two key shortcomings, https and ipv6  - are
> simply two more criteria for the mirror selection, nothing else.  That
> sounds to me like some database changes for storing the mirror
> attributes, plus the appropriate changes to the selection algorithm. 

If we implement it properly, we must regularly probe every mirror for each attribute:

http, https, ipv4 and ipv6 (because certificates are added, expire, network changes, etc).

And then we must track that properly and consider when matching client request.

This is actually what the PR does.

But MirrorBrain also has locking problems when two scanners scan the same folder on the same or differet mirrors, which is hard to track and troubleshoot.

Plus lacks proper WebUI for managing servers and reports, etc.

--
Per Jessen, Zürich (5.8°C)
Member, openSUSE Heroes