| 3 Jul 2025 |
 @saiko:knifepoint.net | every serious network file system that tries to avoid file corruption is going to be slower than sshfs | 10:47:27 |
| @saiko:knifepoint.net | if you’re getting abysmal performance though, make sure it’s not mounted with the sync option though | 10:48:50 |
| @saiko:knifepoint.net | * if you’re getting abysmal performance though, make sure it’s not mounted with the sync option | 10:48:55 |
 K900 | You might also need to enable nfs4.2? | 10:49:27 |
| @saiko:knifepoint.net | I think that should be default? | 10:50:37 |
| @saiko:knifepoint.net | my exports just has /Volumes/storage -rw,sync,no_subtree_check,sec=krb5:krb5i:krb5p,insecure * | 10:50:50 |
| @saiko:knifepoint.net | and the client uses whatever the best is | 10:51:23 |
 @emma:rory.gay | 4.2 feels the same | 10:51:59 |
| @saiko:knifepoint.net | I think it shows the version in the mount command output | 10:52:22 |
| @emma:rory.gay | its not sync, though im trying to avoid extra lookups given that, well... >100ms latency between client and version | 10:52:31 |
| @emma:rory.gay | * its not sync, though im trying to avoid extra lookups given that, well... >100ms latency between client and serve | 10:52:34 |
| @emma:rory.gay | * its not sync, though im trying to avoid extra lookups given that, well... >100ms latency between client and server | 10:52:35 |
| @saiko:knifepoint.net | ah, are you mounting it over the internet? | 10:53:59 |
| @emma:rory.gay | yes, across continents no less | 10:55:22 |
| @emma:rory.gay | (not security sensitive, so lack of encryption is fine) | 10:55:56 |
| @emma:rory.gay | though here's one thing NFS does appear to be far better at: parallel lookups | 10:56:30 |
| @emma:rory.gay | it doesnt do the entire speeding up/hanging/slowing down thing sshfs does | 10:56:44 |
| @saiko:knifepoint.net | okay, no idea about that, I’ve only done NFS across short distances where the latency was somewhat okay, and there it was alright if you were patient | 10:58:23 |
| @saiko:knifepoint.net | maybe a distributed filesystem would be better here if you want this to be fast, though that’s also something I’ve never used (but want to) | 10:59:36 |
| @emma:rory.gay | okay, maybe i should explain what im trying to accomplish in the first place:
im trying to debug and fix some code (nodejs) in production, as it cant really be tested (its not nixos, the src/dist dirs are read/write...)
my IDE really hates remote filesystems, so im trying to make it as smooth as possible (thanks jetbrains)
before even trying to do it this way i was making a git commit, pushing to github and pulling constantly, slowing down any attempt to debug things a lot | 11:02:01 |
| K900 | Jetbrains stuff has proper remoting now no? | 11:02:20 |
| @emma:rory.gay | doesnt that have the entire "oh but my binaries!" problem? | 11:03:27 |
| @saiko:knifepoint.net | can you run it locally and have it access the remote database (especially while the remote is running)? | 11:03:29 |
| @emma:rory.gay | no because the code im debugging has to do with incomming http requests rather than the db/data side of stuff | 11:03:57 |
| @saiko:knifepoint.net | then just push the new build to the remote when it’s fixed | 11:03:59 |
| @saiko:knifepoint.net | ah | 11:04:02 |
| K900 | What problem? | 11:04:08 |
| @emma:rory.gay | just tried to set it up, and yeah it wants to download the entire IDE on my server | 11:06:30 |
| @saiko:knifepoint.net | you could set up the remote web server (probably a reverse proxy?) to route requests over a ssh proxy socket to your computer | 11:06:41 |
| @emma:rory.gay | then im gonna get hyper bottlenecked on the db for sure | 11:07:07 |
