| 26 Jun 2021 |
|  @grahamc:nixos.orgchanged room power levels. | 01:33:14 |
 kevincox | Is there a known issue about systemd ProtectSystem? I have seen a couple of issues about various similar sounding issues. But I am seeing it in a lot of places. For example I currently have to disable it for Tor, systemd-resolved and nginx on nixos-unstable. I'm guessing that something else in my config is breaking it but it seems I am not the only one and I haven't been able to track down the incompatibility. | 14:01:27 |
| kevincox | The error I am getting is:
<pre>Jun 26 12:55:07 t6 systemd[1]: Starting Tor Daemon...
Jun 26 12:55:07 t6 systemd[846]: <font color="#FF7474"><b>tor.service: Failed to set up mount </b></font><span style="background-color:#FF7474"><font color="#0A0713"><b>namespa</b></font></span><font color="#FF7474"><b>cing: /run/systemd/unit-root/: Input/output error</b></font>
Jun 26 12:55:07 t6 systemd[846]: <font color="#FF7474"><b>tor.service: Failed at step NAMESPACE spawning /nix/store/0gh96yhid2i1gks2l5g6silqb61wwy4j-tor-0.4.5.7/bin/tor: Input/output error</b></font>
Jun 26 12:55:07 t6 systemd[1]: <b>tor.service: Control process exited, code=exited, status=226/NAMESPACE</b>
</pre> | 14:24:03 |
| kevincox | * The error I am getting is: | 14:24:31 |
| kevincox | Jun 26 12:55:07 t6 systemd[1]: Starting Tor Daemon...
Jun 26 12:55:07 t6 systemd[846]: tor.service: Failed to set up mount namespacing: /run/systemd/unit-root/: Input/output error
Jun 26 12:55:07 t6 systemd[846]: tor.service: Failed at step NAMESPACE spawning /nix/store/0gh96yhid2i1gks2l5g6silqb61wwy4j-tor-0.4.5.7/bin/tor: Input/output error
Jun 26 12:55:07 t6 systemd[1]: tor.service: Control process exited, code=exited, status=226/NAMESPACE
| 14:24:40 |
| kevincox | * <pre>Jun 26 12:55:07 t6 systemd[1]: Starting Tor Daemon...
Jun 26 12:55:07 t6 systemd[846]: <font color="#FF7474"><b>tor.service: Failed to set up mount </b></font><span style="background-color:#FF7474"><font color="#0A0713"><b>namespacing: /run/systemd/unit-root/: Input/output error</b></font>
Jun 26 12:55:07 t6 systemd[846]: <font color="#FF7474"><b>tor.service: Failed at step NAMESPACE spawning /nix/store/0gh96yhid2i1gks2l5g6silqb61wwy4j-tor-0.4.5.7/bin/tor: Input/output error</b></font>
Jun 26 12:55:07 t6 systemd[1]: <b>tor.service: Control process exited, code=exited, status=226/NAMESPACE</b>
</pre> | 14:31:15 |
| kevincox | * <pre>Jun 26 12:55:07 t6 systemd[1]: Starting Tor Daemon...
Jun 26 12:55:07 t6 systemd[846]: <font color="#FF7474"><b>tor.service: Failed to set up mount namespacing: /run/systemd/unit-root/: Input/output error</b></font>
Jun 26 12:55:07 t6 systemd[846]: <font color="#FF7474"><b>tor.service: Failed at step NAMESPACE spawning /nix/store/0gh96yhid2i1gks2l5g6silqb61wwy4j-tor-0.4.5.7/bin/tor: Input/output error</b></font>
Jun 26 12:55:07 t6 systemd[1]: <b>tor.service: Control process exited, code=exited, status=226/NAMESPACE</b>
</pre> | 14:31:44 |
| 27 Jun 2021 |
|  haugh joined the room. | 13:57:58 |
 Gytis Ivaskevicius | I got a random question:
Do any of you see some init which would seem like a competitor to systemd? | 20:22:23 |
| Gytis Ivaskevicius | The most powerful init i know aside systemd is s6. It has ups and downs compared to systemd but big thing that seems to be lacking is permission control | 20:23:03 |
 Roos | s6 is amazing, but has no real service lifecycle management either IIRC. | 20:24:07 |
| Gytis Ivaskevicius | https://awesome-anti-systemd.netlify.app/ what i find funny that it bearly even manages to fill up the page :D | 20:29:15 |
| 28 Jun 2021 |
 Emil Karlson | it seems to me like systemd competition strives to do nothing new or innovative, but I don't really think mainstream wants to go back to that | 03:43:20 |
 antifuchs | Not many of the things that claim to replace it try to do anything more than be a better init, with maybe some parallelism | 03:48:06 |
| antifuchs | * Not many of the things that claim to replace it try to do anything more than be a better init, with maybe some parallelism and process restarts | 03:48:24 |
| haugh | a lot of the systemd resistance I've seen comes from concern about UNIX philosophy, but (at least from my perspective) putting the whole system layer in one place is the main draw | 03:49:04 |
| haugh | I wish they'd used a standard config format though for crying out loud | 03:49:23 |
| haugh | I really love the mounts and sockets functionality which I haven't found anywhere else. | 03:50:08 |
| antifuchs | Same and also the tempdir & user management | 04:10:09 |
| Gytis Ivaskevicius | In reply to @haugh:matrix.org
a lot of the systemd resistance I've seen comes from concern about UNIX philosophy, but (at least from my perspective) putting the whole system layer in one place is the main draw yeah, i dont really care about that much. If you do - you probably should get rid of stuff like browsers, IDE's, DE's, heck probably Linux kernel as well :D | 04:54:41 |
| Gytis Ivaskevicius | What i do dislike that systemd parts that have value on its own are coupled as part of systemd package and you can not compile them without base init | 04:55:35 |
| Gytis Ivaskevicius | In reply to @haugh:matrix.org
a lot of the systemd resistance I've seen comes from concern about UNIX philosophy, but (at least from my perspective) putting the whole system layer in one place is the main draw * yeah, I don't really care about that much. If someone does wish to follow unix - he probably should get rid of stuff like browsers, IDE's, DE's, heck probably Linux kernel as well :D | 04:56:43 |
| Roos | In reply to @haugh:matrix.org
I wish they'd used a standard config format though for crying out loud It's INI files, is it not? | 05:02:02 |
| haugh | not quite | 05:03:52 |
| haugh | I don't know if this was intentional but the use of a proprietary config format prevents people like me from going whole-hog on generating configs before we understand how the system works. Like instead of generating a bunch of separate service units with heredocs (literally the first thing I did), you should be using transients | 05:06:46 |
| Emil Karlson | transients for what? | 05:08:51 |
| haugh | In reply to @gytis-ivaskevicius:matrix.org
yeah, I don't really care about that much. If someone does wish to follow unix - he probably should get rid of stuff like browsers, IDE's, DE's, heck probably Linux kernel as well :D respectful disagree on this; something like a web browser is a pile of different technologies because you don't control both ends of the conversation. Web design would be somehow even more insane if users could roll their own combinations of JS envs and CSS renderers, for example | 05:09:30 |
| Gytis Ivaskevicius | well in terms of browsers contract is html and css between any two systems | 05:10:46 |
| Gytis Ivaskevicius | as a common language js was accepted and used everywhere | 05:11:00 |
| Gytis Ivaskevicius | but then we have extensions, apps, fancy permission handling, etc | 05:11:19 |
