Jan Pokorný
Jan Pokorný
Note that pacemaker itself also uses `copytruncate` option: https://github.com/ClusterLabs/pacemaker/blob/Pacemaker-1.1.16/extra/logrotate/pacemaker.in#L9 (affects pacemaker prior to switch to corosync's log file in the runtime, or pacemaker_remoted + perhaps more).
One of the things that were brought up in relation this was a question of a backward compatibility, because there's no hard relevance between say pacemaker and its logrotate file...
On 08/12/16 08:14 -0800, Ken Gaillot wrote: >> For the new logrotate config itself and regarding what libqb could be >> taught to do, it was suggested that the libqb...
On 08/12/16 18:57 +0100, Jan Pokorný wrote: > On 08/12/16 08:14 -0800, Ken Gaillot wrote: >>> For the new logrotate config itself and regarding what libqb could be >>> taught...
On 12/12/16 00:22 -0800, Jan Friesse wrote: > My 2cents. Looks like a extremely overengineered solution. Ability > to reopen log seems like a viable feature and it's just good...
I don't know why, I've missed there was a pre-existing issue #142.
Possibly interesting in this context: ~~https://is.muni.cz/th/172673/fi_d/thesis-twoside-final-bw.pdf~~ https://is.muni.cz/th/rjfzq/thesis-twoside-final-bw.pdf
For how Linux deals with this task: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/kernel/printk/printk.c?h=v4.19#n309
See also possible syslog mapping proposed in another context in the past https://cee.mitre.org/language/1.0-beta1/clt.html#clt-syslog-mapping
Ah, even RFC exists (proposed standard, exists since 2009): https://tools.ietf.org/html/rfc5424