bug#32336: herd stop mcron yields instant shutdown of computer

bug#32336: herd stop mcron yields instant shutdown of computer

[Tonton]

[-- Attachment #1: Type: text/plain, Size: 3076 bytes --]

Triggering action:
from bash: sudo herd stop mcron

Result: Exactly the same behaviour as when I call halt/poweroff/reboot from
bash. Except one of the messages clearly seen midscreen on the TTY is
shepherds message that mcron has been stopped. It then scrolls off as the
system is doing it's regular shutdown routine. And reboots, ending at the
normal login screen.

Expected result: mcron service stopped with maybe a message to stdout or log.
System still running.

Guix pull was at:
Generation 9	Jul 28 2018 22:59:26
  guix 48d7ac1
    repository URL: https://git.savannah.gnu.org/git/guix.git
    branch: origin/master
    commit: 48d7ac175f69fea587eaa0358eddb5c76205e8ad

The configuration on the host at the time can be found at:
https://notabug.org/thomassgn/guixsd-configuration/src/e8404bd617547b41d57bceaaa880ca8109360e03/config.scm
(I'm guessing most of it can be stripped out, I'll see if I can reproduce in
a minimal os)

I was implementing the battery-check job for mcron (line 71). Which I know
does not work as is in the config above. It has since been improved.


Here is /var/log/messages around the time (removed ip's and some ntpd lines
because ip's):

750117:Jul 31 18:24:55 localhost shepherd[1]: Service mcron has been started. 
750118:Jul 31 18:24:55 localhost shepherd[1]: Service mcron has been disabled. 
750119:Jul 31 18:24:55 localhost shepherd[1]:   (Respawning too fast.) 
750172:Jul 31 18:37:42 localhost shepherd[1]: Enabled service mcron. 
750173:Jul 31 18:37:42 localhost shepherd[1]: Service mcron has been started. 
750174:Jul 31 18:37:42 localhost shepherd[1]: Respawning mcron. 
750175:Jul 31 18:37:42 localhost shepherd[1]: Service mcron has been started. 
750176:Jul 31 18:37:42 localhost shepherd[1]: Respawning mcron. 
750177:Jul 31 18:37:42 localhost shepherd[1]: Service mcron has been started.  
750178:Jul 31 18:38:20 localhost shepherd[1]: [No documentation.] 
750203:Jul 31 18:43:54 localhost shepherd[1]: Service mcron has been stopped. 
750204:Jul 31 18:43:54 localhost shepherd[1]: Exiting shepherd... 
750205:Jul 31 18:43:54 localhost shepherd[1]: Service console-font-tty2 has been stopped. 
750206:Jul 31 18:43:54 localhost shepherd[1]: Service term-tty2 has been stopped. 
750207:Jul 31 18:43:54 localhost shepherd[1]: Service cups has been stopped. 
750208:Jul 31 18:43:54 localhost shepherd[1]: Service rpcbind-daemon has been stopped. 
750209:Jul 31 18:43:54 localhost avahi-daemon[833]: Got SIGTERM, quitting.
750210:Jul 31 18:43:54 localhost shepherd[1]: Service avahi-daemon has been stopped. 
750211:Jul 31 18:43:54 localhost avahi-daemon[833]: Leaving mDNS multicast group on interface wlp0s20u2.IPv6 with address
750212:Jul 31 18:43:54 localhost avahi-daemon[833]: Leaving mDNS multicast group on interface wlp0s20u2.IPv4 with address
750213:Jul 31 18:43:54 localhost ntpd[827]: ntpd exiting on signal 15 (Terminated)
750218:Jul 31 18:43:54 localhost shepherd[1]: Service ntpd has been stopped. 
750219:Jul 31 18:43:54 localhost connmand[826]: Terminating

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#32336: herd stop mcron yields instant shutdown of computer

[Tonton]

[-- Attachment #1: Type: text/plain, Size: 1267 bytes --]

I attempted to reproduce the bug in a minimal-ish vm. First I ran guix pull
--commit=48d7ac175f69fea587eaa0358eddb5c76205e8ad (the same as the previous
generation where this occured) Then I built the following config as a vm:

(add-to-load-path "/home/ton/guixsd/modules/")
(setenv "GUIX_PACKAGE_PATH" "/home/ton/guixsd/modules/:$GUIX_PACKAGE_PATH")
(use-modules (gnu)
             (gnu system)
             (gnu tests)
             (guix monads)
             (guix store)
             (ice-9 rdelim)
             (srfi srfi-1))

(use-service-modules networking mcron)

(use-package-modules search)

(define updatedb-job
  #~(job '(next-hour '(3))
         (lambda ()
           (execl (string-append #$findutils "/bin/updatedb")
                  "updatedb"
                  "--prunepaths=/tmp /var/tmp /gnu/store"))
         "Updatedb!"))

(define %mcron-os
  (simple-operating-system
   (dhcp-client-service)
   (mcron-service (list updatedb-job))))

%mcron-os

When booted I log in as root, check mcron is running, then stop it. And it
behaves as expected.

So, as of now I have not been able to reproduce my own bug. I experienced it
twice some hours ago, but after running a guix pull and a reconfigure it is
gone AFAIK.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

bug#32336: herd stop mcron yields instant shutdown of computer

[Ludovic Courtès]

Hello,

Tonton <tonton@riseup.net> skribis:

> Triggering action:
> from bash: sudo herd stop mcron

[...]

> 750173:Jul 31 18:37:42 localhost shepherd[1]: Service mcron has been started. 
> 750174:Jul 31 18:37:42 localhost shepherd[1]: Respawning mcron. 
> 750175:Jul 31 18:37:42 localhost shepherd[1]: Service mcron has been started. 
> 750176:Jul 31 18:37:42 localhost shepherd[1]: Respawning mcron. 
> 750177:Jul 31 18:37:42 localhost shepherd[1]: Service mcron has been started.  
> 750178:Jul 31 18:38:20 localhost shepherd[1]: [No documentation.] 
> 750203:Jul 31 18:43:54 localhost shepherd[1]: Service mcron has been stopped. 
> 750204:Jul 31 18:43:54 localhost shepherd[1]: Exiting shepherd... 
> 750205:Jul 31 18:43:54 localhost shepherd[1]: Service console-font-tty2 has been stopped. 
> 750206:Jul 31 18:43:54 localhost shepherd[1]: Service term-tty2 has been stopped. 
> 750207:Jul 31 18:43:54 localhost shepherd[1]: Service cups has been stopped. 
> 750208:Jul 31 18:43:54 localhost shepherd[1]: Service rpcbind-daemon has been stopped. 
> 750209:Jul 31 18:43:54 localhost avahi-daemon[833]: Got SIGTERM, quitting.

The “Exiting shepherd” message above normally only gets printed upon
‘halt’ or ‘reboot’ (or ‘herd stop root’.)  Could it be that one of these
commands was run at that time?

> So, as of now I have not been able to reproduce my own bug. I experienced it
> twice some hours ago, but after running a guix pull and a reconfigure it is
> gone AFAIK.

I’ve just tried in a VM and haven’t been able to reproduce it either.

I’m closing the bug but do reopen it if the problem comes up again.

Thanks,
Ludo’.