Update Debian Package Services and Config Directory by DaAwesomeP · Pull Request #1855 · OpenLightingProject/ola

DaAwesomeP · 2023-04-21T16:45:00Z

Now that I am using the Debian build from master in production I have found some outdated parts.

This updates the Debian package to use a modern Systemd service and moves the config folder. It also removes the reliance on Debconf to enable/disable which definitely not needed for Systemd and is redundant on Debian for init scripts (and especially redundant now that Systemd wraps init scripts by default).

Per Debian package recommendations both the init script and Systemd service will be installed, but users running systemctl start|enable|etc olad will now trigger the Systemd service instead of the init script Systemd wrapper. This fixes a big bug I was experiencing where Systemd does not properly trigger the init script and so systemctl restart olad did not work with the init script but start and stop did.

Systemd users can now override options with Systemd overrides instead of an environment file /etc/default (new preferred method).

Per Debian package recommendations the configuration files should go in /etc/ola since they are user editable and not 100% runtime-controlled. This is also where I think most users expect to find the configuration.

This creates some breaking changes, but I think this is OK for the following reasons:

Until GitHub Actions Debian Builds #1812 the Debian build in this repo was broken, so surely the raw Debian build from this repo has not been used in actual production in some time.
Due to the above I am 99% sure that anyone using OLA on a Debian-based system is either:
a. Building from source, installing with make install, and creating their own config directory and service scheme
b. Installing from the Debian package repository which already uses /etc/ola for configs and has removed the Debconf bits
I am submitting this against master instead of 0.10.

These changes should make it much smoother to move between the Debian repo packages and the ones created here. As new features and bugfixes seem to have accelerated recently, the ability to use the pre-packaged master build is quite nice. Personally I experienced this because I needed KiNETv2 support (#1787).

The main remaining difference in the packages is in the way that the web assets are handled, but most users won't perceive a difference between these builds with that or would possibly try this build if that one breaks. I don't really see a need to update that part here.

Hopefully this eventually makes its way downstream to Debian and brings the Systemd service to everyone.

Co-authored-by: Wouter Verhelst <w@uter.be> Co-authored-by: Perry Naseck <git@perrynaseck.com>

DaAwesomeP · 2023-04-21T21:31:22Z

Requesting review from @peternewman

DaAwesomeP · 2023-04-30T21:17:22Z

OK, I have this running on a production machine and it works like a charm!

DaAwesomeP · 2023-05-17T20:15:38Z

@peternewman polite ping!

yoe

I know you didn't ask for my review, but I'm giving it anyway ;-P

Although I made some suggestions for improvement, none of them are really critical, so from my POV things can go in as-is and that would be better than the status quo; but I'll leave actual approval up to @peternewman.

DaAwesomeP · 2023-05-19T13:17:22Z

I know you didn't ask for my review, but I'm giving it anyway ;-P

I didn't ask but you're an infinitely more experienced package maintainer than me and I am excited that you left a review! :D

peternewman · 2023-06-21T03:02:34Z

Note from the init/systemd side there's also #1444 which I suspect is stuck on me looking at some stuff (as always)...

…-systemd-and-config-dir

…:DaAwesomeP/ola into DaAwesomeP-debian-systemd-and-config-dir

DaAwesomeP · 2026-03-21T17:48:17Z

OK, I have rebased and the build is working for Debian bullseye, bookworm, and trixie. Debian forky (testing) and sid (unstable) are showing C++ compilation errors to be fixed in other PRs:

./include/ola/rdm/ResponderTagSet.h:191:15: error: 'uint8_t' has not been declared
./include/ola/rdm/ResponderTagSet.h:195:7: error: 'uint8_t' was not declared in this scope
./include/ola/rdm/ResponderTagSet.h:195:16: error: 'ptr' was not declared in this scope
common/rdm/ResponderTagSetTest.cpp:56:29: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'
common/rdm/ResponderTagSetTest.cpp:83:30: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'
common/rdm/ResponderTagSetTest.cpp:84:29: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'
common/rdm/ResponderTagSetTest.cpp:100:29: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'

…-systemd-and-config-dir

aroffringa

I went over the changes and it looks great to me. It's hard to tell everything is robust to the future but I think that's fine, we can deal with that later. Shall I merge it?

DaAwesomeP · 2026-04-04T22:00:31Z

I went over the changes and it looks great to me. It's hard to tell everything is robust to the future but I think that's fine, we can deal with that later. Shall I merge it?

It's really up to you! I've been ready for a while :D. @yoe approved it (and handles package management on the Debian side of things). I don't know that @peternewman ever reviewed it.

…-systemd-and-config-dir

DaAwesomeP · 2026-04-04T22:01:33Z

Just rebased again.

peternewman

Thanks again @DaAwesomeP . A few relatively minor comments from me...

peternewman · 2026-04-04T23:35:59Z

@@ -0,0 +1,16 @@
+[Unit]
+Description=Open Lighting Architecture daemon
+Documentation=man:olad(1)


I'd be keen to add these too:
https://github.com/OpenLightingProject/ola/pull/1444/files#diff-f6474409cdd97933e66d7f3eb687aa1bc0c23b9a6785394a5a1eb884a00d7d29R3-R5

Can we add these in a followup PR? I want to avoid tacking on anything else to this PR since it has been open so long.

Separately, I think these links should go in the man page instead of in the Systemd service. That way you see them no matter how you get to the docs on Debian. Also means fewer places to keep them updated.

Can we add these in a followup PR? I want to avoid tacking on anything else to this PR since it has been open so long.

Yeah fair point.

Separately, I think these links should go in the man page instead of in the Systemd service. That way you see them no matter how you get to the docs on Debian. Also means fewer places to keep them updated.

I think they should perhaps go in both (assuming people actually put web links in man pages, although it feels a bit like defeating the point of man. I'd hope we never need to update them anywhere regardless, as we'd just put redirects in.

peternewman · 2026-04-04T23:44:43Z

We should also consider what other settings we should merge in from https://github.com/OpenLightingProject/ola/pull/1444/files#diff-f6474409cdd97933e66d7f3eb687aa1bc0c23b9a6785394a5a1eb884a00d7d29 initially, versus possibly wait for the notify stuff in the next release?

This PR has been open for nearly 3 years, so I'd like to just get this going and come back to the other parts.

Also discussed this here in this PR: #1855 (comment)

Co-authored-by: Peter Newman <peternewman@users.noreply.github.com>

peternewman

LGTM, lets finally get this in...

peternewman · 2026-04-18T16:02:56Z

+DAEMON_ARGS="--world-writable"

 # Reads config file (will override defaults above)
 [ -r /etc/default/ola-rdm-tests ] && . /etc/default/ola-rdm-tests

-if [ "$RUN_DAEMON" = "true" ] || [ "$RUN_DAEMON" = "yes" ] ; then
-  DAEMON_ARGS="--world-writeable"


After a bit of headbanging at Plugfest (not helped by SystemD silently stealing control of the init script code, I eventually spotted we'd changed to the other spelling of the argument (without the e).

For consistence (for now) I've corrected it in:
b860a75

I guess we could think about a breaking change to drop the e at some point in the future @DaAwesomeP ?

I eventually spotted we'd changed to the other spelling of the argument (without the e).
I guess we could think about a breaking change to drop the e at some point in the future @DaAwesomeP ?

Oh wow huh. I probably did this without realizing. I definitely did not mean to change the spelling of any existing args.

After a bit of headbanging at Plugfest (not helped by SystemD silently stealing control of the init script code

I will say, this has been the default behavior for a very long time now. I was going to suggest fully removing the init script in v0.12.0 or later.

Is there a reason you were still using the init script over the Systemd one? You should be able to trivially set OLAD_OPTS in a Systemd override. As simple as: systemctl edit olad and Environment=OLAD_OPTS=myopts.

I eventually spotted we'd changed to the other spelling of the argument (without the e).
I guess we could think about a breaking change to drop the e at some point in the future @DaAwesomeP ?

Oh wow huh. I probably did this without realizing. I definitely did not mean to change the spelling of any existing args.

No need to apologise, all three of us reviewing it missed it too!

After a bit of headbanging at Plugfest (not helped by SystemD silently stealing control of the init script code

I will say, this has been the default behavior for a very long time now. I was going to suggest fully removing the init script in v0.12.0 or later.

Ah right, I didn't know. Wouldn't that break things for people choosing not to use SystemD (I've no idea how many people that actually is).

Is there a reason you were still using the init script over the Systemd one? You should be able to trivially set OLAD_OPTS in a Systemd override. As simple as: systemctl edit olad and Environment=OLAD_OPTS=myopts.

My issue was SystemD wasn't telling me what was wrong. It was only when I ran with set -x and stopped the include function dragging in SystemD so it really ran my bash code, I actually saw the error onscreen to realise what needed fixing. Also I've rarely had to override things at all, let alone in SystemD (does that also mean it won't load the /etc/defaults file too anymore)? Anyway I don't know if that's an us issue, our SystemD config, or SystemD itself, but I never saw the error in journalctl, just that it had exited.

peternewman · 2026-04-18T16:18:02Z

OK, I have rebased and the build is working for Debian bullseye, bookworm, and trixie. Debian forky (testing) and sid (unstable) are showing C++ compilation errors to be fixed in other PRs:

./include/ola/rdm/ResponderTagSet.h:191:15: error: 'uint8_t' has not been declared

./include/ola/rdm/ResponderTagSet.h:195:7: error: 'uint8_t' was not declared in this scope

./include/ola/rdm/ResponderTagSet.h:195:16: error: 'ptr' was not declared in this scope

common/rdm/ResponderTagSetTest.cpp:56:29: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'

common/rdm/ResponderTagSetTest.cpp:83:30: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'

common/rdm/ResponderTagSetTest.cpp:84:29: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'

common/rdm/ResponderTagSetTest.cpp:100:29: error: cannot convert 'uint8_t*' {aka 'unsigned char*'} to 'int*'

It would be good to know what's still broken after various other stuff has gone in now @DaAwesomeP ?

DaAwesomeP · 2026-04-20T01:54:14Z

It would be good to know what's still broken after various other stuff has gone in now @DaAwesomeP ?

It looks like tests are building and passing in master so we must be good?

peternewman · 2026-04-25T10:16:40Z

It looks like tests are building and passing in master so we must be good?

Yeah, I'm just a bit confused given I don't think I did anything to fix the last few...

DaAwesomeP and others added 14 commits April 20, 2023 15:22

debian add initial systemd service

9d6b770

debian/ola.postinst: add --home parameter to adduser invocation

693dc63

Co-authored-by: Wouter Verhelst <w@uter.be> Co-authored-by: Perry Naseck <git@perrynaseck.com>

debian remove debconf from init script, add OLAD_OPTS

14d6d7e

debian move config to /etc/ola

2758515

Don't use debconf for this

48d7dfb

More debconf removals

12090e6

Co-authored-by: Wouter Verhelst <w@uter.be> Co-authored-by: Perry Naseck <git@perrynaseck.com>

add rdm_test_server systemd service

84282f4

debian systemd services user, rdm_test_server service configurable args

9ffb314

debian systemd services wait for network, olad

896b13f

update Debian compat level

bd89102

debian dh-autoreconf superseeds autotools-dev

e905f02

debian rules remove autotools_dev but keep package

d24b096

Debian stick with debhelper 12

886c092

Debian add override_dh_installsystemd

24d2cec

yoe reviewed May 19, 2023

View reviewed changes

Comment thread debian/changelog

Comment thread debian/control

DaAwesomeP added 3 commits June 18, 2023 16:25

Merge branch 'master' into DaAwesomeP-debian-systemd-and-config-dir

977dca8

debian update Makefile dist files list

b31073a

debian update Makefile dist files list again

146a9d1

DaAwesomeP added 7 commits June 22, 2023 16:42

Merge remote-tracking branch 'upstream/master' into DaAwesomeP-debian…

8729000

…-systemd-and-config-dir

add build name generation and bake into build

1740d5a

Merge branch 'DaAwesomeP-debian-systemd-and-config-dir' of github.com…

0bf40e7

…:DaAwesomeP/ola into DaAwesomeP-debian-systemd-and-config-dir

build name move git check to script (can be used outside of autoconf)

b66a17a

github actions debian use git describe for version

f903797

github actions debian git build name fix env assignment

938733c

github actions debian git build name fix env assignment typo

5111428

CI add Debian forky to debian-test

ccb128e

DaAwesomeP mentioned this pull request Mar 30, 2026

Fix Sid dependencies and autopkgtest #2033

Merged

Merge remote-tracking branch 'upstream/master' into DaAwesomeP-debian…

30906cf

…-systemd-and-config-dir

DaAwesomeP added this to the 0.11.0 milestone Apr 4, 2026

DaAwesomeP assigned aroffringa Apr 4, 2026

DaAwesomeP added the enhancement label Apr 4, 2026

DaAwesomeP requested a review from aroffringa April 4, 2026 19:16

aroffringa approved these changes Apr 4, 2026

View reviewed changes

Merge remote-tracking branch 'upstream/master' into DaAwesomeP-debian…

c00183a

…-systemd-and-config-dir

DaAwesomeP enabled auto-merge April 4, 2026 22:11

DaAwesomeP disabled auto-merge April 4, 2026 22:12

peternewman requested changes Apr 4, 2026

View reviewed changes

Update debian/ola-rdm-tests.rdm_test_server.service

08d6a95

Co-authored-by: Peter Newman <peternewman@users.noreply.github.com>

DaAwesomeP requested a review from peternewman April 5, 2026 00:26

DaAwesomeP added Maintainability OpSys-Linux labels Apr 5, 2026

DaAwesomeP mentioned this pull request Apr 5, 2026

v0.11.0 #2038

Open

26 tasks

peternewman mentioned this pull request Apr 6, 2026

Master resync #2040

Merged

peternewman approved these changes Apr 6, 2026

View reviewed changes

peternewman merged commit 6e8147a into OpenLightingProject:master Apr 6, 2026
7 of 24 checks passed

peternewman reviewed Apr 18, 2026

View reviewed changes

This comment was marked as off-topic.

Sign in to view

peternewman mentioned this pull request Jul 12, 2026

Add 0.11 news #2060

Draft

Uh oh!

Conversation

DaAwesomeP commented Apr 21, 2023

Uh oh!

DaAwesomeP commented Apr 21, 2023

Uh oh!

DaAwesomeP commented Apr 30, 2023

Uh oh!

DaAwesomeP commented May 17, 2023

Uh oh!

yoe left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

DaAwesomeP commented May 19, 2023

Uh oh!

peternewman commented Jun 21, 2023

Uh oh!

DaAwesomeP commented Mar 21, 2026

Uh oh!

aroffringa left a comment

Choose a reason for hiding this comment

Uh oh!

DaAwesomeP commented Apr 4, 2026

Uh oh!

DaAwesomeP commented Apr 4, 2026

Uh oh!

peternewman left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

peternewman left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

This comment was marked as off-topic.

This comment was marked as off-topic.

peternewman commented Apr 18, 2026

Uh oh!

DaAwesomeP commented Apr 20, 2026

Uh oh!

peternewman commented Apr 25, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

5 participants