"GOT", but the "O" is a cute, smiling pufferfish. Index | Thread | Search

From:
Stefan Sperling <stsp@stsp.name>
Subject:
Re: got fetch downloads too much
To:
Christian Weisgerber <naddy@mips.inka.de>
Cc:
gameoftrees@openbsd.org
Date:
Wed, 6 Oct 2021 18:23:16 +0200

Download raw body.

Thread
On Wed, Oct 06, 2021 at 05:22:20PM +0200, Christian Weisgerber wrote:
> For 0.61, stsp fixed a bug where "got fetch" downloaded too many
> objects, but this just caught my eye:
> 
> Connecting to "origin" anongit@git.freebsd.org
> server: Enumerating objects: 82587, done.
> server: Counting objects: 100% (19234/19234), done.
> server: Compressing objects: 100% (12280/12280), done.
> server: Total 82587 (delta 13996), reused 6954 (delta 6954), pack-reused 63353
>     75M fetched; indexing 100%; resolving deltas 100%
> Fetched a849279233753372e82e4784ff444c0071eeb8e2.pack
> Updated refs/remotes/origin/main: da3278ded3b2647d26da26788bab8363e502a144
> Updated refs/remotes/origin/stable/12: fb49012ca448f3c7a9ae38ceaf700b8cd39be836
> 
> Something isn't right here.  I had fetched a few hours earlier, and
> in the intervening time 4 commits were added to main and 1 to
> stable/12, none of them large.  75M is four orders of magnitude too
> much.

Is it reproducible? You could try "rewinding" those refs to the
commits which you started fetching with, e.g. with 'got ref'.
And then fetch again. Does it still fetch more than it should?
If that works, we would have an easy test case.