On 6/6/20 7:05 PM, Tyler Dence wrote:
I (arlm.tyzoid.com) would prefer an opt-in solution. Such a method would allow us to evaluate in advance the storage requirements to make sure that we don't run out of storage space on the virtual disk.
How it's implemented with rsync is an open question. Would it exist as a new endpoint? If so, wouldn't it necessitate extra config on the apache/nginx side?
On Sat, Jun 6, 2020, 6:37 PM Tails Hon1nbo <hon1nbo+mirror@hackingand.coffee> wrote:
Hacking & Coffee would be interested, but make it a separate "opt-in" dataset especially if it's unpredictable how much space it will take. If suddenly enabled, and set to opt-out, I imagine some mirror clusters would have a sudden, unexpected storage problem The storage requirements should not sharply rise, unless we do a mass rebuild that enables debug packages. I think a reasonable thing to do from the packaging side of things is to rebuild some critical libraries, but leave most packages to simply acquire debug packages whenever they are rebuilt for other reasons.
We could also give advance notice, e.g. something along the lines of "we will enable debug packages next month, make sure your disks can handle irregular growth of up to XXX gb or set the following rsync exclusion in advance and re-evaluate after the churn". Would this be a reasonable way to handle it? It would have the advantage of not requiring new rsync endpoints and server configs. -- Eli Schwartz Bug Wrangler and Trusted User