Re: Anubis check stuck?

Le 06/08/2026 à 21:12, Mark Wielaard a écrit :
> Hi David,
> 
> On Thu, Aug 06, 2026 at 08:36:09AM +0100, David Abdurachmanov wrote:
>> On Wed, Aug 5, 2026 at 7:53 PM Alice Carlotti via Gcc <[email protected]> wrote:
>>>
>>> I have just had this issue with patchwork.sourceware.org (stuck in a hash then
>>> reload loop) and gcc.gnu.org/bugzilla (no visible activity on anubis display; I
>>> think this affected gcc.gnu.org/git/gitweb.cgi as well).
>>
>> I noticed this, and decided to try these links on Firefox and
>> Chromium. Especially with Chromium as I never used it to access these
>> URLs.
>>
>> Patchwork got stuck with "Calculating...". Twice, even on Incognito mode.
> 
> OK. After consultation with Xe I updated all Anubis environments with
> COOKIE_PREFIX=swanubis0806. That should resolve the hickups with
> existing cookies. But also invalidates everyones existing cookies.
> 
> For new Anubis releases Xe will make sure updated cookies shouldn't be
> an issue anymore.
> 
>> gcc.gnu.org/git/gitweb.cgi or https://sourceware.org/git/?p=glibc.git
>> is basically impossible in general. It's "429 Too Many Requests" for
>> most of the week in general. You pass Anubis and instantly hit with
>> 429.
> 
> So this isn't a specific Anubis issue. It is the new nginx rate
> limiter. If your ip is "authenticated" (for example after login to
> bugzilla) you won't get rate limited. Maybe we need more ways to
> authenticate?
> 
> I assume you are seeing this with gitweb because it is making too many
> requests. Have you tried cgit? https://sourceware.org/cgit/ That
> should be more efficient and hopefully doesn't get you rate limited
> that quickly.

Hello,

I have been seeing the 429 error recently using unauthenticated git 
(command line, not any web page frontend) from the compile farm 
machines.  Is it possible to whitelist the machines or is the preferred 
way to use git+ssh or some other authentication mechanism?