worldofpeace_ changed the topic of #nixos-dev to: #nixos-dev NixOS Development (#nixos for questions) | NixOS stable: 20.03 ✨ | | | 19.09 RMs: disasm, sphalerite; 20.03: worldofpeace, disasm |
alp has quit [Ping timeout: 264 seconds]
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-dev
<julm> damned, using a /nix/store over a ZFS dataset using normalization=formD breaks the libsass unpacking, which contains two directories whose names have the same normalization :(
<gchristensen> ouch.
<julm> wget -O - | tar -tzf - | grep unicode-pwd
<julm> there is already an extraPostFetch=''rm -r $out/test/e2e/unicode-pwd'' to fix the build on HFS+, but on ZFS it fails before that, when moving from the tmpfs to the ZFS
<julm> mv: inter-device move failed: '/build/unpack/libsass-3.6.3/test/e2e/unicode-pwd/S'$'\303\241''ss-U'$'\305\242''F8' to '/nix/store/bqfq4db6nwycmkdrql9igsbrayqsw3g2-source/test/e2e/unicode-pwd/S'$'\303\241''ss-U'$'\305\242''F8'; unable to remove target: Directory not empty
<gchristensen> honestly don't know what to say or do :
teto has quit [Ping timeout: 240 seconds]
<{^_^}> #86432 (by ju1m, 47 seconds ago, open): libsass: fix unpacking on ZFS Nix store with normalization=formD
<emily> you might want to consider disabling filename unicode normalization, since it usually only breaks stuff
<emily> (I guess this is on macOS though?)
<cole-h> What's the invocation to rebase onto staging properly?
<samueldr> cole-h: rebase on what git merge-base origin/staging origin/master tells you IIRC
delroth has quit [Changing host]
delroth has joined #nixos-dev
<cole-h> Thanks. veprbl got to it before me :P I'll add that to my growing nixos notes, though.
drakonis has quit [Quit: WeeChat 2.8]
cole-h has quit [Quit: Goodbye]
alp has joined #nixos-dev
<lovesegfault> gchristensen: is stateless NixOps in the cards?
<srk> lovesegfault: it is - storage.memory
<{^_^}> nixops#1264 (by grahamc, 5 weeks ago, open): Example NixOps State Backends
FRidh has joined #nixos-dev
phreedom has quit [Ping timeout: 240 seconds]
<lovesegfault> Oh, nice, I had seen that but wasn't sure whether it was sufficient
phreedom has joined #nixos-dev
<lovesegfault> srk: what about rootless NixOps :P
<srk> lovesegfault: not sure what do you mean by that
<lovesegfault> it seems like it tries to ssh in as root
<lovesegfault> as opposed to ssh'ing as my user and escalating to root with sudo
<srk> ah, that's just a default I would say, there's a PR for escalation as well by adisbladis IIRC
<lovesegfault> Oh, I see it
<{^_^}> nixops#1270 (by adisbladis, 4 weeks ago, open): Add support for non-root deployments
<{^_^}> nixops#1283 (by adisbladis, 2 weeks ago, open): Add SSH jump host support
<{^_^}> nixops#1264 (by grahamc, 5 weeks ago, open): Example NixOps State Backends
<lovesegfault> with those three I can start using NixOps :D
<lovesegfault> So close!
<srk> :D
regnat[m] has joined #nixos-dev
alp has quit [Remote host closed the connection]
orivej has quit [Ping timeout: 256 seconds]
justanotheruser has quit [Ping timeout: 272 seconds]
alp has joined #nixos-dev
__Sander__ has joined #nixos-dev
__Sander__ has quit [Ping timeout: 246 seconds]
__monty__ has joined #nixos-dev
primeos has quit [Quit: WeeChat 2.8]
<bennofs[m]> Where's the current staging jobset on Hydra? The ones in nixos/staging and nixpkgs/staging got evaluated in 2019 last
<bennofs[m]> Ah, staging-next is probably what I'm looking for
justanotheruser has joined #nixos-dev
<manveru> any ideas why i get `error: store path name is empty` when using `nix-store`?
<clever> manveru: what cmd did you run to get the error?
<manveru> for example `sudo nix-store --verify`
<manveru> i think something is wrong with my nix db...
<clever> manveru: yeah, one min...
<clever> manveru: `sqlite3 /nix/var/nix/db/db.sqlite .dump` would dump the entire thing as sql, then you could just pastebin the thing
<manveru> hm
<manveru> that funnily also fails :)
<clever> manveru: definitely sounding like your db is corrupted
* srk thought sqlite was hard to corrupt
<clever> yeah, it should be pretty hard to mess up
<manveru> well, i can copy the db and dump it
<manveru> just dumping directly gives this
<manveru> ah, for some reason that also needs root
<clever> manveru: it might be trying to auto-repair the db, and it needs write access to do so
prusnak has quit [Changing host]
prusnak has joined #nixos-dev
prusnak has joined #nixos-dev
<manveru> well, i did a .backup now
<manveru> backup also doesn't work, neither does recover
<manveru> i think there must be some invalid value in the db, not a corruption
<manveru> interestingly that error seems to come from the rust part of nix
<clever> manveru: if not even sqlite3 can .dump, then its definitely corruption
<clever> manveru: when things are that far gone, the simplest thing you can do is just blow away the entire /nix and re-run nixos-install to recreate it
<manveru> i can dump a copy of the db
<manveru> just that making a new db from that dump gives the same failure
<clever> manveru: ah, can you pastebin the full dump?
<manveru> it's like 300MB... so no
<clever> yikes
<manveru> i looked through the db, there are no empty paths
<manveru> but ok, will just recreate from scratch if that's no issue :)
<clever> yeah, thats in the rust code, ive not read this area yet...
<clever> it can occur if a StorePathName is constructed with an empty string
<clever> manveru: are you able to get a backtrace with gdb?
nschoe has joined #nixos-dev
alp has quit [Quit: Leaving]
<niksnut> manveru: try this: sqlite3 /nix/var/nix/db/db.sqlite 'delete from validpaths where length(path) = 44'
<niksnut> probably you should make a backup of db.sqlite first :p
<manveru> i have backups :)
alp has joined #nixos-dev
<manveru> niksnut: that seems to work
<clever> oh, so it was a path in the form of /nix/store/9krlzvny65gdc8s7kpb6lkx8cd02c25b-
<clever> no name, but it does have a hash
<manveru> yep
<manveru> that was the invalid entry
<manveru> this only happened after upgrading to nixFlakes and trying a GC, given that the registration date is 2019-11-26
__Sander__ has joined #nixos-dev
<julm> emily: unfortunately normalization is readonly, it can only be set when creating the pool. I was new to ZFS when creating that pool, and only followed the recommendation on which is normalization=formD
<qyliss> It's not really a recommendation
<qyliss> It just says you can
<gchristensen> guh, deleting that, it doesn't make sense to recommend things just because
<gchristensen> s/recommend/mention/
<gchristensen> julm: I was tempted by that one too
<emily> julm: ah, I see. yeah that's a terrible recommendation
<emily> could probably use someone going over and updating it (e.g. I guess zstd might be better than lz4 for new pools now?)
<gchristensen> emily: really it shouldn't recommend any specific algo, just "on" which picks the best available
<emily> on macos/windows it makes some sense to mangle filenames because the APIs are based around unicode, but linux is pretty firmly in the bag of "filenames are raw byte sequences"
<gchristensen> emily: I've tried cleaning it up a few times, but kept getting bogged down infeeling like there was so much to do and not feeling good about where to start. ... so I just wrote this:
<emily> gchristensen: sure, as long as on picks good values (zstd has adjustable compression level I guess?)
<emily> I just have some huge zpool/zfs commands commented out in my config for reference
<emily> of course as luck would have it those are out of date now too
<gchristensen> yea
<julm> qyliss: yes, mention, but this appears with only a warning for the adventuruous for a whole list of options which are not adventuruous at all to me like compression or even required like acltype=posixacl, so I used it :|
<gchristensen> (zstd isn't in 20.03, so not sure it sohuld be recommended)
<gchristensen> julm: yeah, the others really are recommendations
<gchristensen> julm: I'm sorry :(
<emily> gchristensen: fwiw just replacing the article with a better one would be welcome, I think, as long as it contains all the info you need to set up the basic setups people want
<julm> gchristensen: don't be, it's not a big deal :]
<emily> I did that for (the dnscrypt page was stuck on dnscrypt-proxy1, the DNS over TLS one recommended dnsmasq)
<emily> of course i guess ideally this stuff would go in the manual
<emily> once it's not docbook I guess :p
<qyliss> Yeah, ZFS in the manual would be great
<gchristensen> +1
<julm> there is a departure from docbook on the way for the manual?
<gchristensen> there is an RFC
FRidh has quit [Quit: Konversation terminated!]
<{^_^}> rfcs#64 (by Infinisil, 16 weeks ago, open): [RFC 0064] New Documentation Format for nixpkgs and NixOS
<adisbladis> [ZFS in manual]: is also pretty good stuff
<adisbladis> etu++
<{^_^}> etu's karma got increased to 15
<emily> TIL boot.loader.grub.mirroredBoots
<emily> does systemd-boot have that?
<emily> nope :(
<emily> that really has no reason to be in the grub module
<adisbladis> I think not?
<adisbladis> Is it even possible to have mirrored ESPs?
<adisbladis> Oh right, I misunderstood how mirroredBoots work
<emily> yes, that's what the option does :)
<emily> you just make two /boots
<emily> and ensure they're both in the UEFI variables
<emily> if the boot of the first one fails, it falls back to the second
<emily> (same as how USB drive boot falls back to disk or whatever)
<adisbladis> emily: Then mirroredBoots should totally be moved out of grub :)
<emily> gchristensen: dang, "this is not exactly how my set up mine due to my usage of tmpfs as root, but that's a topic for another post" is a real tease there
<emily> adisbladis: it does look like the other bootloaders would need some adaptation of it, since it just installs grub twice rather than copying the files
<adisbladis> emily: Me and etu use a very similar setup with tmpfs /
<emily> (which um, confuses me, since I'm not sure how that would copy the kernels/initrds)
<qyliss> I've been thinking about a tmpfs / on the new computer
<adisbladis> The gist is: /nix on persistent storage (zfs), stuff that you want persistent are symlinked to persistent storage
orivej has joined #nixos-dev
alp has quit [Ping timeout: 240 seconds]
<adisbladis> Even for $HOME
<gchristensen> emily: did you see the next post?
<emily> I admit that this penguin beanie is probably better than any /-on-tmpfs explanation could possibly be.
* adisbladis <- proud owner of that beanie
teto has joined #nixos-dev
<etu> emily, adisbladis: Mirrored boots is amazing though :)
<etu> Super useful feature
<aanderse> qyliss: i just built a new computer and did / on tmpfs... but then failed and gave up halfway through, so /etc is also mounted to persistent storage :\
<aanderse> if i get some energy i might try and troubleshoot that at some point
obadz- has joined #nixos-dev
obadz has quit [Ping timeout: 264 seconds]
obadz- is now known as obadz
<emily> is there a nice way to see what stuff in /etc isn't tracked by nixos?
<emily> I guess looking for non-symlinks is a first approximation
<adisbladis> emily: Approximation being the operative word
<adisbladis> Some files are created by systemd units
<emily> yeah. I guess you could diff /etc against /etc/static?
<emily> adisbladis: rather few, it seems
<emily> I guess I'm relatively hygienic already :)
dongcarl has joined #nixos-dev
alp has joined #nixos-dev
cole-h has joined #nixos-dev
v0|d has quit [Remote host closed the connection]
justanotheruser has quit [Ping timeout: 240 seconds]
justanotheruser has joined #nixos-dev
nschoe has quit [Ping timeout: 240 seconds]
nschoe has joined #nixos-dev
drakonis has joined #nixos-dev
primeos has joined #nixos-dev
alp has quit [Ping timeout: 272 seconds]
alp has joined #nixos-dev
alp has quit [Ping timeout: 240 seconds]
alp has joined #nixos-dev
<talyz> aanderse: To make tmpfs on / easier for me, I created two modules, one for NixOS one for home-manager
<aanderse> thanks talyz I'll take a look!
justanotheruser has quit [Ping timeout: 265 seconds]
__monty__ has quit [Quit: leaving]
__Sander__ has quit [Ping timeout: 258 seconds]
zarel_ has joined #nixos-dev
zarel has quit [Ping timeout: 265 seconds]
Scriptkiddi has quit [*.net *.split]
ajs124 has quit [*.net *.split]
avn has quit [*.net *.split]
niksnut has quit [*.net *.split]
primeos has quit [*.net *.split]
andi- has quit [*.net *.split]
NinjaTrappeur has quit [*.net *.split]
tdeo has quit [*.net *.split]
aminechikhaoui has quit [*.net *.split]
qyliss has quit [*.net *.split]
klys has quit [*.net *.split]
kgz has quit [*.net *.split]
aranea has quit [*.net *.split]
{^_^} has quit [*.net *.split]
lopsided98 has quit [*.net *.split]
greizgh has quit [*.net *.split]
thonkpod has quit [*.net *.split]
niksnut has joined #nixos-dev
avn has joined #nixos-dev
andi- has joined #nixos-dev
NinjaTrappeur has joined #nixos-dev
andi- has joined #nixos-dev
andi- has quit [Changing host]
tdeo has joined #nixos-dev
thonkpod has joined #nixos-dev
primeos has joined #nixos-dev
Scriptkiddi has joined #nixos-dev
tdeo has joined #nixos-dev
tdeo has quit [Changing host]
primeos is now known as Guest90765
qyliss has joined #nixos-dev
greizgh has joined #nixos-dev
kgz has joined #nixos-dev
lopsided98 has joined #nixos-dev
klys has joined #nixos-dev
Guest90765 has quit [Quit: WeeChat 2.8]
primeos_ has joined #nixos-dev
notgne2 has quit [Quit: ZNC 1.7.4 -]
justanotheruser has joined #nixos-dev
notgne2 has joined #nixos-dev
primeos_ has quit [Quit: WeeChat 2.8]
primeos_ has joined #nixos-dev
primeos_ is now known as primeos
aminechikhaoui has joined #nixos-dev
<worldofpeace> Jan Tojnar: was the giDiscoverSelf fix backported?
<jtojnar> worldofpeace no, I thought the regression was only on unstable
<worldofpeace> Jan Tojnar: I believe we backported the gappsWrapperArgsHook change
<jtojnar> oh, then this would need backporting too
<jtojnar> I think I only checked what branches the commit is on, which would not show cherry-picks
{^_^} has joined #nixos-dev
alp has quit [Ping timeout: 244 seconds]
alp has joined #nixos-dev
justanotheruser has quit [Ping timeout: 260 seconds]
justanotheruser has joined #nixos-dev