Hey,

I've seen the gitlab issue as well, I'll try to find what could be the issue. As Jeremy pointed out in the issue it could be mirrors that sync often so that could in turn mean that when a sync is executed the check from archweb is as well and that could cause some extra latency for the check to fail. This is my very short guess but i'll try to dig in more.

Regards,
Arun

On 1/29/24 02:25, JING LUO wrote:

Hello all,

(Jeremy Kescher: if you send html emails, could you please use a text
color that is not white please :)

On Mon, Jan 29, 2024 at 7:38 AM Jeremy Kescher <jeremy@kescher.at> wrote:

No, system time is synced everywhere, including on the mirror servers. So that probably doesn't have to do with anything.

On Sun, 2024-01-28 at 20:18 +0000, archlinux.rs738@simplelogin.com wrote:

Could be completely irrelevant Jeremy however, is your system's time & date gone out of sync?


Regards,
Liam O'Luachra





-------- Original Message --------
On 29 Jan 2024, 05:21, Jeremy Kescher < jeremy_at_kescher_at_wqykx@simplelogin.co> wrote:


I run the mirrors at https://archlinux.org/mirrors/kescher.at/ and unfortunately, I keep seeing a rather rare "Could not parse time from lastsync" on the Arch mirror status site, curiously on both rsync and https. This prevents completion from reaching 100%, which is rather annoying to me, and presumably some folks that depend on the mirror's existence. I have my own monitoring set up to check lastsync, and I have no recorded failure. Same goes for error logs I have set up specifically to debug this. Does anyone here have experience in debugging this? I have no idea what could be causing this anymore.

It's happening to my mirror https://archlinux.org/mirrors/jing.rocks/
and many others too, and my server recorded no sync failure.

Please also see the gitlab issue I opened last week:
https://gitlab.archlinux.org/archlinux/arch-mirrors/-/issues/52#note_159516

It's happening more often than I thought it was.

Hi, as I've posted on GitLab: I've talked to DevOps team and it might be an issue where due to increaisng IO writing the lastsync at T0 might take some time, causing an actually empty lastsync file to be served, causing the checks to fail as well. Hopefully this change will fix that infrastructure!794

Regards,
Arun