Hi team,
My mirror was pulled off from the Fedora mirrormanager few months ago without notice, I reviewed and found that it actually happened across all projects using mirrormanager v2.
There is an issue with mirrormanager that it considers "HTTP/3 alt-svc upgrade header returned via HTTP" invalid. Returning that header via HTTP basically instructs all clients to initiate UDP connection on a specific port if possible, which isn't wrong since clients can still choose to not follow. In my case, the HTTP/3 is deployed on port 443 (because QUIC is encrypted by default), which essentially makes the HTTP request (via HTTP/3) HTTPS always.
I have no idea why mirrormanager insists on following the upgrade header, but because I will not turn off HTTP/3, HTTP mirror URLs will be removed.
After I removed all HTTP links from the site, Fedora mirrormanager still has not crawled my site again since June.
Can you check what is still preventing the crawling there? I would avoid deleting the entire mirror site and create a new mirror.
Thanks!
Is there a reason this needs to be private?
CC: @abompard
Metadata Update from @cicku: - Issue private status set to: False (was: True)
Metadata Update from @phsmoura: - Issue priority set to: Waiting on Assignee (was: Needs Review) - Issue tagged with: low-gain, low-trouble, ops
Hello! I looked at the site in mirrormanager, it's been disabled by an admin with the comment: "Reaching out to the mirror admin for rsync whitelist". Does this ring a bell?
No, that is my own comment since the inception of the mirror site, never changed since 2024.
it's been disabled by an admin with the comment
I don't see such? <img alt="Screenshot_2025-10-27_114802.png" src="/fedora-infrastructure/issue/raw/files/63733138a0de3992e27d8efd8275b66c4069d47b9fcd4c1ea800ecc222ec5156-Screenshot_2025-10-27_114802.png" />
Yeah it's an admin flag that only admins see. I've re-enabled the site now.
Metadata Update from @kevin: - Issue assigned to kevin
I see crawl logs now, so I assume this is all done? If anything feel free to reopen and let us know.
Metadata Update from @kevin: - Issue close_status updated to: Fixed - Issue status updated to: Closed (was: Open)
Log in to comment on this ticket.