| holmanb | rbasak: Reseting machine-id at runtime would force boot order requirements on all services (present and future) that use machine-id. Do we have an idea of which/how many services this might affect? | 01:31 |
|---|---|---|
| lunatiq | in regard to postfix/mailutils to enable mail() on dedicated machine. Do I select Internet Site for each domain or do I create a domain for it? | 01:46 |
| lunatiq | can I just use local and use it for local use only? | 01:47 |
| lunatiq | Would that enable scripts on local machine to use mail() ? | 01:47 |
| blahdeblah_ | lunatiq: If all you want is mail on local machine then you can choose local | 01:59 |
| lunatiq | blahdeblah_ does that mean PHP can use mail() function? | 03:48 |
| blahdeblah_ | lunatiq: I don't know enough about the PHP mail() function to answer that, but if it can be configured to use localhost as its SMTP server, or /usr/bin/mail as its local mailer, then I'd think it would work. | 03:55 |
| === remolej3 is now known as remolej | ||
| === elastic_dog is now known as Guest823 | ||
| effendy[m] | holmanb: systemctl -t service | nc -l for instance? Services is a broad word. | 08:23 |
| effendy[m] | broad idea* | 08:23 |
| === ajfriesen6 is now known as ajfriesen | ||
| === diddledani is now known as Guest9829 | ||
| === diddledani_ is now known as diddledani | ||
| === diddledani is now known as diddledani_ | ||
| === diddledani_ is now known as diddledani | ||
| Odd_Bloke | holmanb: I guess my response would be that those services doing the wrong thing earlier in boot (due to a stale machine-id) is not a desirable outcome, so it seems irrelevant how many/which services would be affected. | 16:09 |
| Odd_Bloke | Having said that, I don't see a systemd level dependency on /etc/machine-id in any units on my system, so I don't know if there will be an _easy_ way to do that. | 16:13 |
| Odd_Bloke | s/do that/determine that set of services/ | 16:13 |
| Odd_Bloke | As a data point, DigitalOcean already have vendor-data to handle resetting machine-id on first boot, because not doing so causes problems. | 16:14 |
| holmanb | Odd_Bloke: Which services seems relevant in that any that are ordered before cloud-init would have to be restarted with the new machine-id - and really any that use it should probably be explicitly ordered after whichever cloud-init unit would be made to do the runtime rewriting of machine-id. | 16:18 |
| holmanb | Odd_Bloke: How many services was just a question of the scope that such a change would affect. | 16:19 |
| Odd_Bloke | holmanb: Ack, understood, thanks! | 16:22 |
| ahasenack | rbasak: did we ever plan or think about extracting detached orig tarball signatures, if they exist in lp or pristine-tar, when running git ubuntu export-orig ? | 17:33 |
| rbasak | ahasenack: I was not aware that they exist | 19:37 |
| ahasenack | I don't know if they are handled by pristine-tar (i.e., ingested), or LP | 19:38 |
| ahasenack | but every now and then I see a dpkg warning during build that a gpg key exists in the source, but no detached signature | 19:38 |
| ahasenack | and I think this could be improved, if the .sig files are stored somewhere | 19:38 |
| rbasak | Sounds reasonable, if they work everywhere else through the pipeline | 19:38 |
| ahasenack | dpkg-buildpackage seems to check it when building it seems | 19:39 |
| ahasenack | but I haven't verified. I just saw the warning | 19:39 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!