8 Nov 2018 |
_hc | adding disk is easy, but it is a support request, so it can take days til its done | 11:49:09 |
Mathijs | eh, I'd script a warning message if the disk starts running low | 11:49:39 |
Coffee | AIUI you can't use half of your disk space if you want to be able to vacuum your database. | 11:49:49 |
Mathijs | that should give us plenty of time | 11:49:54 |
Nico | Mathijs: Do you think we can make a machine which has public logs and an IRC bridge beside Synapse? The latter one is most important of course, the formers would be cool, too, to depend even less on matrix.org. | 11:50:30 |
Coffee | So how does 100G disk sound? | 11:50:37 |
Mathijs | 100GB wouls sound enough | 11:51:02 |
_hc | sounds like too much to me for this | 11:51:15 |
Coffee | Nico: I'd be willing to set up a botbot if someone can provide the necessary infrastructure. | 11:51:16 |
_hc | that's a lot | 11:51:21 |
Mathijs | In reply to @nico:dorfbrunnen.eu Mathijs: Do you think we can make a machine which has public logs and an IRC bridge beside Synapse? The latter one is most important of course, the formers would be cool, too, to depend even less on matrix.org. public logs: sure, we'd have to set up whatever matrix is running for the static view of the room, or even write a bot | 11:51:33 |
_hc | that'll be the second biggest VM after the mirror | 11:51:42 |
Mathijs | but the irc bridge is a bit trickier | 11:51:44 |
Mathijs | in the sense that we'd need freenode to allow the bridge, I think | 11:51:59 |
Coffee | In reply to @eighthave:matrix.org that'll be the second biggest VM after the mirror We could probably go with 50G if we're more careful. | 11:52:51 |
_hc | 100GB of storage that also needs to be backed up | 11:52:59 |
Mathijs | 50 sounds doable | 11:53:02 |
_hc | sounds like | 11:53:05 |
Coffee | But yeah, synapse is a disk hog. | 11:53:10 |
Mathijs | yeah, synapse likes it disk usage | 11:53:20 |
Nico |
in the sense that we'd need freenode to allow the bridge, I think
OK, so that's to be done later. | 11:53:38 |
Mathijs | if the DB is on spinning rust, we'll also probably notice a bit of slowness | 11:53:52 |
Mathijs | In reply to @nico:dorfbrunnen.eu
in the sense that we'd need freenode to allow the bridge, I think
OK, so that's to be done later. I'm not 100% that we'd technically need them to allow the bridge, but I do know that matrix.org has an agreement with freenode, and if nothing else it's the polite thing to do | 11:54:33 |
Nico | Couldn't we reduce the needed disk space if we regularly clear the cache of loaded media? Of course it should automatically reload the media if it's requested. | 11:54:33 |
Mathijs | yeah, that could work | 11:55:40 |
Mathijs | though in my case the database is larger than my media cache | 11:55:53 |
Coffee | In reply to @nico:dorfbrunnen.eu Couldn't we reduce the needed disk space if we regularly clear the cache of loaded media? Of course it should automatically reload the media if it's requested. Not that much. Most of the storage goes to Matrix "state". | 11:56:16 |
Coffee | Our best bet is to just keep the number of rooms low. We can start with just our two F-Droid rooms. | 11:57:05 |
Coffee | And anyone who gets an account on it has to pledge to not join any other rooms than those two. | 11:57:31 |
Marcus (old) | In reply to @eighthave:matrix.org we should meet up 😀 Definitely! | 11:57:38 |