Control: forwarded -1 
https://public-inbox.org/git/20200422084254.GA27502@furthur.local/

On 2020-03-23 18:57 +0100, Sven Joachim wrote:

> On 2020-03-23 07:10 +0100, Sven Joachim wrote:
>
>> Package: git
>> Version: 1:2.26.0~rc2-1
>> Severity: important
>>
>> I have a git repository for the Linux kernel where I track both Linus'
>> and the stable repository.
>>
>> ,----
>> | $ git remote --verbose show
>> | origin  git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 
>> (fetch)
>> | origin  git://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git 
>> (push)
>> | stable  https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git 
>> (fetch)
>> | stable  https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git 
>> (push)
>> `----
>>
>> A few days ago, after the update to 2.26.0-rc2, I fetched from the
>> stable tree.  This triggered a run of "git gc --auto" in the background
>> as there were more than 6700 loose objects.
>>
>> This morning I pulled from Linus' tree, and it downloaded _a lot_ of
>> objects.  Both that and resolving deltas took quite some time, as one
>> might imagine.
>>
>> ,----
>> | $ git pull
>> | remote: Enumerating objects: 7226920, done.
>> | remote: Counting objects: 100% (7226920/7226920), done.
>> | remote: Compressing objects: 100% (1098637/1098637), done.
>> | remote: Total 7226920 (delta 6082531), reused 7224317 (delta 6080838)
>> | Receiving objects: 100% (7226920/7226920), 1.18 GiB | 3.95 MiB/s, done.
>> | Resolving deltas: 100% (6082531/6082531), done.
>> `----
>
> Additionally, the size of the .git directory ballooned to ~3.5 GiB after
> that.  Running "git gc" manually reduced it to ~2 GiB which I think is
> normal.

It seems this has been reported upstream in the meantime, and the fix
(or workaround) was to disable protocol version 2 by default in git
2.27.

Meanwhile, after seeing the problem again I had switched to https for
Linus' repository which also helped.

I think the bug should be either closed or the severity downgraded, but
I leave the decision to the package maintainer who also participated in
the upstream discussion. :-)

Cheers,
       Sven

Reply via email to