[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Large data packages in the archive

Luk Claes <luk@debian.org> writes:

> Raphael Geissert wrote:
>> Hi all,
>> What about going the 'b.)' way but define it as a RG (or even RC) with some
>> other changes to policy (like requiring big data package's source packages
>> to be arch-indep and not build anything else but the data packages). 
>> That way the transition could be done gradually for lenny+1 so there's no
>> bloating.
>> And, mirror admins could then have plenty of time to decide whether to
>> mirror the data packages or not.
> Are you sure that the current sync scripts make that possible and won't
> sync everything unless explicitely stated differently and will keep
> working without intervention for the time being? Because otherwise it's
> like Joerg said not an option IMHO.

It is quite simple to ignore a suite in the official mirror scripts
and they will just sync everything without intervention.

I also think that any space pressed mirror admin has no problems
adding an exclude.

What I think could be problematic is that mirrors are chained. If any
mirror in the middle of the chain excludes the data suite then all
mirrors downstream will do so too. I'm not sure how long the chains
are but there could be surprising effects.

Excluding the data suite could only be allowed if all downstream
mirrors agree they don't want that or get changed to a different

Overall it is nothing unsurmountable but it would certainly cause some
disruptions. Hopefully this showed you some of the risk of option b.


Reply to: