periklis has quit [Remote host closed the connection]
LnL has quit [Ping timeout: 245 seconds]
LnL has joined #nixos-security
erictapen has joined #nixos-security
tokudan[m] has joined #nixos-security
erictapen has quit [Ping timeout: 246 seconds]
__Sander__ has joined #nixos-security
LnL has quit [Ping timeout: 252 seconds]
LnL has joined #nixos-security
LnL has quit [Ping timeout: 252 seconds]
LnL has joined #nixos-security
__Sander__ has quit [Quit: Konversation terminated!]
<andi->
So there appears to be some fallout from the systemd patches... surprise ;-) The fsck tests aren't passing anymore and potentially other stuff. Seen on 18.09 backport. I am trying to look into the fsck thing. The test is also rather recent might have had false assumptions before...
<pie__>
andi-, nixos tests catching things? :D
<pie__>
great :>
<andi->
I am a bit confused where that expected After and requires lines were being generated
<andi->
interestingly systemd is generating broken symlinks on both the old and the new version -.-
<gchristensen>
:|
<andi->
so the fstab generator didn't change but doesn't emit two lines anymore
<andi->
so running the generator again after the system booted yields the correct results.. This smells..
<pie__>
ah the age old, run it again and reboot till it works:>
pie__ is now known as pie_
<andi->
well that solve the issue.. I have just redirected the output to /tmp/foo
<andi->
and there is appears to be correct.. I am trying to find the place that actually calls the generator
MichaelRaskin has joined #nixos-security
<andi->
wow, systemd checks if it can find a `fsck.${fstype}` executable during boot in it's PATH. If that isn't found then there is no fsck unit entry for that thing... so why did this break?
<andi->
the good thing is modifying stage1 / stage2 for debugging is super easy with nixos... the good & bad thing is that we must be strict with PATH and must propagated that properly..
<andi->
so this was introduced with 2f05cd12dd8f3c27f69681afb4c9f0f1883a9a6c & probably made a bug with dd3be5ae941617f70bf1fc1cb21776e242c58cd0
<andi->
(a bug for us probably, not in general)
erictapen has joined #nixos-security
LnL has quit [Ping timeout: 245 seconds]
LnL has joined #nixos-security
<andi->
Sorry for the wall of text, it miused this as a personal log /o\
<andi->
yeah that fixes it.. one more world rebuild /o\
<fpletz>
andi-: awesome, I've been going through the queue of backports for 18.09, some of them are also mass rebuilds
<fpletz>
let's merge all of them with your systemd fix \o/
<andi->
Investigating the hibernate test failure next
* andi-
wonders how that hibernate test worked with 1MB of storage o.O
<andi->
should get some sleep.. playing with disks and vm test drivers leaks old disks.. :/
<andi->
6789dca0a26df0c44ff8020f0a4206bf21e52a7a broke hibernation since we never configured it and it probably just worked by chance?!?
<andi->
adding `boot.kernelParams = [ "resume" ];` to the hibernate test makes it work.. not sure how we should handle that and how that should automagically happend for users that had/have a working resume configuration :/
<andi->
time to head home and switch devices... will open those PRs as soon as I am home...
<andi->
chromium probably will still block the channel bump because >=70 has a different sandbox output..