4 Oct 2024 |
mjm | and this is solving a pretty edge case in impermanence i think | 03:21:48 |
ElvishJerricco | but sure | 03:21:49 |
mjm | idk how i'd do it | 03:22:14 |
mjm | i think this wouldn't be an issue for LUKS, since the decrypted FS has its own device and would show up separately from the encrypted one | 03:23:26 |
ElvishJerricco | My first thought was to make a service for the directory you want to make that has RequiresMountsFor=/sysroot/persist/whatever and Before=/sysroot/bind/whatever that does the mkdir | 03:23:56 |
ElvishJerricco | that way you don't have to make these assumptions they're making | 03:25:12 |
mjm | yeah it's weird to me that they would mount it somewhere temporary instead of just introducing a service after /persist or whatever mounts and before the bind mounts | 03:26:23 |
mjm | https://github.com/nix-community/impermanence/issues/222 if you feel like chiming in | 03:27:27 |
ElvishJerricco | mjm: I mean frankly I think the problem that the service is trying to solve is a silly one | 03:28:47 |
ElvishJerricco | like yes, you should make your directories | 03:28:54 |
ElvishJerricco | it's weird that impermanence has to automated it for you | 03:29:07 |
mjm | yeah, i think maybe it enables bootstrapping a system with impermanence from the get-go? idk i saw mention of nixos-anywhere in the PR that introduced it | 03:29:36 |
mjm | but like, this hasn't been there for a while and i never noticed | 03:29:57 |
mjm | so 🤷 | 03:30:06 |
mjm | anyway, the fstab generator is pretty slick, excited to have it working! | 03:31:22 |
ElvishJerricco | yea, thanks for testing! | 03:35:39 |
aloisw | In reply to @elvishjerricco:matrix.org doesn't btrfs handle large numbers of snapshots absolutely terribly or something? The snapshots themselves are not so much the problem, it's mostly the combination with quota that causes the performance issues. | 12:56:08 |
5 Oct 2024 |
| jeroen left the room. | 13:57:48 |
| magic_rb changed their profile picture. | 22:17:59 |
6 Oct 2024 |
steveej | i'm trying to set up ipv6 prefix delegation with systemd-networkd on my router. i also run dnsmasq there and i'm not sure how relevant that is.
what i'm wondering is how i can confirm that my router is actually getting a delegated prefix from my ISP's router, or whether it just gets a single IP6 address | 09:35:49 |
steveej | i believe i set all the necessary options on the router's WAN interface to accept RAs and also use dhcpv6 to ask for a delegated prefix. | 09:36:31 |
K900 | ip a should show a prefix on the interface | 09:37:05 |
K900 | If you're getting a prefix | 09:37:12 |
steveej | do you mind showing me an example output for that? this is the wan interface output
4: wan@eth0: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 state UP qlen 1000
inet6 2001:XXX:9d94:4420:d4ac:76ff:fec0:7837/64 scope global dynamic mngtmpaddr noprefixroute
valid_lft 7149sec preferred_lft 3549sec
inet6 fdaa:XXX:ddee:0:d4ac:76ff:fec0:7837/64 scope global mngtmpaddr noprefixroute
valid_lft forever preferred_lft forever
inet6 fe80::d4ac:76ff:fec0:7837/64 scope link proto kernel_ll
valid_lft forever preferred_lft forever
slightly paranoid so i XXXd some values there
| 09:38:42 |
K900 | You have a /64 | 09:39:17 |
K900 | So yes that's a prefix | 09:39:21 |
K900 | But normally you'd get a /56 from your ISP | 09:40:01 |
K900 | For prefix delegation to work correctly | 09:40:05 |
K900 | You can configure it to delegate smaller prefixes | 09:40:18 |
K900 | But a typical setup would be to have a /48 or /56 for your router, and a /64 per device | 09:41:09 |