[Piuparts-devel] restarting slaves individually

[Andreas Beckmann]

On 05/06/2019 02.49, Andreas Beckmann wrote:
> 22542e959 p-s: re-exec on SIGUSR1, picking up updated code and new config sections
> 3e9142135 slave-bin/slave_reexec: new, send SIGUSR1 to all running slaves
> 
> stopping the running slaves may take a long time ... making restarts take a long time,
> too.
> Instead have the slaves listen on SIGUSR1 and re-exec themselves as early as possible
> after receiving the signal - after finishing the current test and flushing logs.
> 
> 28817e6c7 network_issues: relax the 'Package .* .* not found in .*' pattern
> 86818aa39 rotate_master_logs: delete master logs older than 90 days
> c6d2062d4 detect_piuparts_issues: clean up stale temporary and empty files
> 07a1b2c31 add logratate exception for chef-server-api
> 1f06f6053 redo the kFreeBSD-only exception
> c30d9400b add skip_distupgrade_from checking the install candidate version
> c95599eb7 enigmail/jessie is not installable with thunderbird 60
> f96b12702 xul-ext-sogo-connector/jessie is not installable with thunderbird 60
> 66abf2c25 xul-ext-compactheader/wheezy,jessie is not installable with thunderbird 60

> 2ec34c945 do not torture *-dkms packages with ancient kernel headers

removed the last one (needs reworking) and added three more:

4515a1251 several firefox,tunderbird/wheezy,jessie cruft packages are uninstallable
25ef98a12 lib/db: factor out common log processing code
c8ff41cc5 lib/db: remove stale .kpr files after receiving new logs

Andreas