Ludovic Courtès <[email protected]> writes: > Hi Tomas, > > Tomas Volf <[email protected]> skribis: > >>> Namely, ‘log-rotation’ is more opinionated and simpler than rottlog in >>> several ways: there’s only a global configuration (rather than >>> per-file), only one file (re)naming scheme, and probably a couple >>> more. >> >> 1. Per-log file settings. Some files I want to never rotate (the data >> has sentimental value). Some files I want to rotate, but keep *all* >> previous versions (if I would be using Guix at work, this is required >> for auditing reasons). >> >> 2. Option to limit number of rotated versions being kept. Currently, >> any single "log stream" can occupy unlimited disk space even with >> frequent rotations. To prevent that, I would like to express expiry >> in terms of "keep current and 4 rotated versions", but that is not >> possible as far as I can tell. >> >> 3. Run-time discovery of log files to rotate. I have no idea how to use >> shepherd for rotating samba's log files for example, their names are >> (in my configuration) dynamic based on the host name of the machine >> that connects to the samba share. Their only shared trait is that >> they are under /var/log/samba.d and end in .log. > > On topic #3, I’m soliciting opinions on this change: > > https://codeberg.org/shepherd/shepherd/pulls/143 > > TIA. :-)
I mostly read the changes to the documentation, and it looks good. I need to read up more about the `@quotation Warning' section, I wonder how (if?) traditional log rotation programs handle this problem. Thanks for letting me know about this, Tomas -- There are only two hard things in Computer Science: cache invalidation, naming things and off-by-one errors.
