Devuan bug report logs - #882
/etc/logrotate.d/lighttpd: Questionable use of invoke-rc.d

Reply or subscribe to this bug.

Toggle useless messages

Message #5 received at submit@bugs.devuan.org (full text, mbox, reply):

Package: lighttpd
Version: 1.4.69-1

/etc/logrotate.d/lighttpd uses invoke-rc.d to notify lighttpd to
reopen logs but it doesn't work when using
policy-rcd-declarative-deny-all so shouldn't the logrotate
configuration snippet use rather 'service' ?

Regards,
Opty

Message #8 received at 882@bugs.devuan.org (full text, mbox, reply):

Control: tags -1 debian

Opty,

On Tue, Apr 01, 2025 at 02:41:25PM +0200, Opty wrote:
> Package: lighttpd
> Version: 1.4.69-1
> 
> /etc/logrotate.d/lighttpd uses invoke-rc.d to notify lighttpd to
> reopen logs but it doesn't work when using
> policy-rcd-declarative-deny-all so shouldn't the logrotate
> configuration snippet use rather 'service' ?

Possibly, but I am not sure. Certainly all the other /etc/logrotate.d examples I
can find use invoke-rc.d. It may be to prevent rotation in chroots and the like
when it would fail.

However, I am sure that lighttpd isn't forked in Devuan and so this should be
reported directly to Debian's BTS.

Thanks

Mark

Message #15 received at 882@bugs.devuan.org (full text, mbox, reply):

On Tue, Apr 1, 2025 at 6:36 PM Mark Hindley <mark@hindley.org.uk> wrote:
> However, I am sure that lighttpd isn't forked in Devuan and so this should be
> reported directly to Debian's BTS.

Now that Debian uses systemd by default I considered Devuan a better
place for System V style init stuff but I will think about reporting
to Debian BTS then.

Regards,
Opty

Message #20 received at 882@bugs.devuan.org (full text, mbox, reply):

On Wed, 2 Apr 2025 13:56:10 +0200
Opty <opty77@gmail.com> wrote:

> On Tue, Apr 1, 2025 at 6:36 PM Mark Hindley <mark@hindley.org.uk>
> wrote:
> > However, I am sure that lighttpd isn't forked in Devuan and so this
> > should be reported directly to Debian's BTS.
> 
> Now that Debian uses systemd by default I considered Devuan a better
> place for System V style init stuff but I will think about reporting
> to Debian BTS then.

Please don't: while this may solve your use case it will break other
more common use cases that are currently working.
Example:
lighttpd installed but disabled by the local admin (K links) but if
service wrapper is used instead of invoke-rc.d the logrotate snippet
will try to signal lighttpd regardless, which should not happen.

on a side note, runit is currently using invoke-rc.d to forward
logrotate actions to runit services, so your change would break that too

Regards,
Lorenzo

> 
> Regards,
> Opty
> _______________________________________________
> devuan-dev internal mailing list
> devuan-dev@lists.dyne.org
> Manage your subscription:
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/devuan-dev
> Archive: https://lists.dyne.org/lurker/list/devuan-dev.en.html

Message #25 received at 882@bugs.devuan.org (full text, mbox, reply):

On Wed, Apr 2, 2025 at 2:12 PM Lorenzo <plorenzo@disroot.org> wrote:
> On Wed, 2 Apr 2025 13:56:10 +0200
> Opty <opty77@gmail.com> wrote:
> > Now that Debian uses systemd by default I considered Devuan a better
> > place for System V style init stuff but I will think about reporting
> > to Debian BTS then.
>
> Please don't: while this may solve your use case it will break other
> more common use cases that are currently working.
> Example:
> lighttpd installed but disabled by the local admin (K links) but if
> service wrapper is used instead of invoke-rc.d the logrotate snippet
> will try to signal lighttpd regardless, which should not happen.

Good, no work for me. :-)

I should have written that I was actually afraid of something like
that in general but 'service' seemed as the best proposal at that
time. See also #881.

Regards,
Opty

Message #30 received at 882@bugs.devuan.org (full text, mbox, reply):

On Wed, Apr 2, 2025 at 2:42 PM Opty <opty77@gmail.com> wrote:
> On Wed, Apr 2, 2025 at 2:12 PM Lorenzo <plorenzo@disroot.org> wrote:
> > Please don't: while this may solve your use case it will break other
> > more common use cases that are currently working.
> > Example:
> > lighttpd installed but disabled by the local admin (K links) but if
> > service wrapper is used instead of invoke-rc.d the logrotate snippet
> > will try to signal lighttpd regardless, which should not happen.
>
> Good, no work for me. :-)
>
> I should have written that I was actually afraid of something like
> that in general but 'service' seemed as the best proposal at that
> time. See also #881.

I recalled why I proposed 'service' and even reported all of these:

"invoke-rc.d and policy-rc.d are precisely interfaces expected to be
used only by maintainer scripts. Anything else done during run-time
should be using some other interface, such as service(8) for example
or an init specific one, otherwise that should be considered a bug."

and

"In case this is a local problem where you have deployed
non-maintainer scripts using invoke-rc.d, then while you fix those,
you could check in policy-rc.d if you are running inside dpkg by
checking if the environment variable DPKG_RUNNING_VERSION is defined
(documented in dpkg(1))."

(see https://bugs.debian.org/804018#25)

Regards,
Opty

Message #35 received at 882@bugs.devuan.org (full text, mbox, reply):

Control: forcemerge 881 -1

On Wed, Apr 02, 2025 at 01:58:20PM +0200, Opty wrote:
> Please see #881 or should I copy the story here as well?

No. Really these are all the same issue. I will merge them for now while we look
at the detail.

Mark

Send a report that this bug log contains spam.