On Tue, Jul 29, 2025 at 4:54 AM Neal Gompa <ngomp...@gmail.com> wrote: > > On Tue, Jul 29, 2025 at 3:46 AM Zdenek Dohnal <zdoh...@redhat.com> wrote: > > > > Hi, > > > > per > > https://fedoraproject.org/wiki/Changes/FoomaticRipRejectsUnknownValues > > I've got feedback about turning RPM scriptlet into systemd service to > > have the change applied even on immutable systems, like bootable > > containers, Fedora CoreOS etc. > > > > Does anyone use such systemd service for upgrade tasks in their packages? > > > > Would you mind sharing the name of component which uses such service, so > > I can check how it is supposed to look? > > > > In general, it's a flawed way to approach the problem. > > If it was trivial to handle, it would probably be done by the programs > themselves automatically. If it is not trivial to handle, it is not > reasonable to use a systemd service to do it. >
That is a surprisingly naive thing to hear you say, because I know you've been around for a long time. Yes, it's *ideal* when the application itself handles config migration automatically, but it's hardly common. Most applications just ship a release note and usually an updated default configuration and that's it. > Delayed execution of scriptlets by migrating them to systemd units > also creates problems because it de-links the execution from the > transaction. This means that the risk of breakage goes up depending on > things happening on the system between transaction execution and the > reboot. It also means that the risk of breakage becomes limited to the single application instead of risking aborting a DNF transaction and leaving the system in an unbootable state. I'm also not sure what you mean by " the risk of breakage goes up depending on things happening on the system between transaction execution and the reboot." unless you mean that people are writing these systemd units to try to fire after the transaction completes (in the same boot) instead of making the new upgrade unit a prerequisite for starting the service the next time it gets loaded. Or are you worried about the user performing manual edits during this timeframe? I'm not really sure what situation you're concerned about here. > My previous attempts at doing stuff like this (like rpmdb-rebuild and > rpmdb-migrate) wound up being fraught with peril (and my fixes never > were merged, resulting in somewhat broken migrations for years). If > something needs to be executed as part of the transaction, it just > needs to happen there. If it can be done outside of the transaction, > it should be handled as part of the program or service itself, and not > as a separate systemd unit. -- _______________________________________________ devel mailing list -- devel@lists.fedoraproject.org To unsubscribe send an email to devel-le...@lists.fedoraproject.org Fedora Code of Conduct: https://docs.fedoraproject.org/en-US/project/code-of-conduct/ List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines List Archives: https://lists.fedoraproject.org/archives/list/devel@lists.fedoraproject.org Do not reply to spam, report it: https://pagure.io/fedora-infrastructure/new_issue
