bug#56519: Shepherd non-deterministically fails to load the "guix-daemon" service after the "user-processes" service has been started

[Markus Nilsson]

Hi Ludo,

Ludovic Courtès <ludo@gnu.org> writes:

>Could you check what /var/log/messages says for this specific occurrence?

>It will shed some light as to what services were started, in what order, how they failed, etc.

I have attached the kernel log information for today (August 5 2022) as well as the output of "herd status". You can see that on line 1020 "user-processes" successfully starts, but "guix-daemon" or "nscd" doesn't attempt to start afterwards for some unknown reason. Below is a summary of the relevant events from the log:

1 - System starts booting

873 - "user-processes" service fails to start

908 - "user-processes" service fails to start a 2nd time

910 - "nscd" service fails to start

920 - "user-processes" service fails to start a 3rd time

922 - "guix-daemon" service fails to start

943 - "user-processes" service fails to start a 4th time

1020 - "user-processes" service starts successfully

1071 - "ntpd" service starts successfully

1075 - "mcron" service starts successfully

This seems to indicate that the Shepherd will attempt to start "user-processes" multiple times until it succeeds. But Shepherd won't try to start "guix-daemon" or "nscd" if it has failed once.

Cheers

Markus

Aug_5_kernel_log
Description: Binary data

Aug_5_herd_status
Description: Binary data