Hey,

In summary: data.guix.gnu.org is going to be down while I attempt to
restore a backup from earlier this week.

I noticed today that this revision [1] hasn't been processed
correctly. One indication of that is the number of packages (9058) is
much less than you'd expect.

There was another revision being processed at the same time [2], and it
seems like the recent revision (33d2574) has been mixed up with this
previous revision.

1: https://data.guix.gnu.org/revision/33d2574d77f36e05065b5fd70cbbd9eb1fa128f3
2: https://data.guix.gnu.org/revision/684626763fa1fc66f4d7f7d909bf85a52d769b02

I can imagine how this isn't impossible, the channels code in Guix uses
a cached checkout, and this would have been at the older commit for
revision [2] before it should have switched to 33d2574. I've got no idea
how this failure could arise though, I've attached the log and there's
no indication there that something went wrong, apart from the wrong data
seeming to have been used.

So, there's two problems here, the wrong data in the data.guix.gnu.org
database, and this mysterious issue that caused it. To address the first
problem, my plan is to restore a backup taken late on Monday (13th).

Unfortunately this will mean data.guix.gnu.org will be down until that
backup is restored, and then it'll take a little while to catch up. This
will affect dependent things like bordeaux.guix.gnu.org, which will lag
behind until data.guix.gnu.org has caught up.

This could be an issue limited to the Guix Data Service, but the code
here is basically the same as is used in guix pull/guix time-machine, so
if if anyone has seen something like this, or has any ideas about what
might have caused this, just let me know.

Thanks,

Chris