<gchristensen> https://gist.github.com/grahamc/9aa8186180507a14e7e9d3f3b27f2202 anyone wnt to take a look at this boot log and guess why it is stalling?
<samueldr> gchristensen: fdisk being the last executed command is not related, is it?
<gchristensen> I...don't think so: this is what is rnning: https://github.com/nix-community/aarch64-build-box/blob/master/persistence.sh#L9
<gchristensen> this works properly on the aarch64.nixos.community box but not this other one
<samueldr> asking since people are seemingly having issues with the manual and fdisk commands
<samueldr> yeah, if it works on the other one :/
<gchristensen> this particular branch hasn't run in "a while" but it certainly did run in the past, successfully
<samueldr> same nixpkgs revisions? (not sure if relevant)
<gchristensen> definitely not
<gchristensen> this is that armv7/aarch64 box sphalerite was alluding to earlier
<gchristensen> it never got this far before, always crashed earlier in the boot
<samueldr> oh! I thought this was the aarch64 builder
<samueldr> so, same hardware as the community box, etc
<gchristensen> it isn't
<gchristensen> :)
* samueldr hadn't realised it wasn't 96 cpu in the long list of cpus
<gchristensen> _only_ 64
<samueldr> instead of fdisking, if you were to sleep, and print the date like once a sec or once 10/60 sec, would it also fail?
samrose_ has quit [Quit: Page closed]
<samueldr> my question: is it the last action that causes a failure, or is stage-1 somewhat unrelated?
<gchristensen> a good question
<samueldr> if stage-1 is unrelated, an assumption could be the kernel is having a bad day
<gchristensen> a decent assumption
<samueldr> 4.17 isn't the latest stable, and 4.19 is soon going to be stable
<gchristensen> +for i in `seq 1 120`; do
<gchristensen> + date
<gchristensen> + sleep 1
<gchristensen> +done
<gchristensen> if ! test -b /dev/sda1; then
<gchristensen> - sed -e 's/\s*\([\+0-9a-zA-Z]*\).*/\1/' << EOF | fdisk /dev/sda
<gchristensen> + sed -e 's/\s*\([\+0-9a-zA-Z]*\).*/\1/' << EOF | strace fdisk /dev/sda
<samueldr> at least sygin isn't here :)
<samueldr> good thinking strace
<gchristensen> I'm sygin proof :)
<gchristensen> since {^_^} operates as my IRC user
<gchristensen> oh right, since the builder is dead (see nixos-dev) I can't build a new image to test.
<thefloweringash> sphalerite: which microSD card did you go with? I have yet to find one that's not disappointing
<thefloweringash> switching to f2fs definitely helps, but it can still stall
<{^_^}> scaleway/image-alpine#51 (by CensoredUsername, 1 year ago, open): No entropy available for /dev/random and getrandom() on a base install
<samueldr> ooooh
<gchristensen> I guess cat'ing some static junk to /dev/random is a bad idea :P
<samueldr> isn't it one-way?
<samueldr> though, helps reproducibility!
<gchristensen> I think knowing the starting state of the randomness is bad
<samueldr> afaiui, it sounds about right
<samueldr> though I think /dev/random isn't writeable?
* samueldr is mildly annoyed at his allwinner board
<samueldr> the ethernet was working... until I rebooted
<gchristensen> it is actually
<gchristensen> Dezgeg: maybe you can take a look at this? https://gist.github.com/grahamc/3a9e6b13df8ae44fdf2fb79db0985be5
orivej has quit [Ping timeout: 252 seconds]
<sphalerite> gchristensen: I think the appropriate solution to the randomness problem is putting haveged in the initrd and starting it before you need the randomness
<sphalerite> but it's not randomness in this case is it?
<sphalerite> or rather lack of randomness
<sphalerite> thefloweringash: samsung evo 128GB, supposedly U3 speed class
<thefloweringash> I have the evo+ 64gb. Let me know how it goes!
<sphalerite> oh yeah might have been evo+
<clever> Model Number: Samsung SSD 960 EVO 500GB
<clever> Total NVM Capacity: 500,107,862,016 [500 GB]
<clever> from my laptop
<thefloweringash> I wish I had the choice of a real storage device and not a microSD card
Thra11 has joined #nixos-aarch64
Thra11 has quit [Ping timeout: 245 seconds]
Acou_Bass has quit [Ping timeout: 260 seconds]
<Dezgeg> gchristensen: yeah I have seen that. maybe with luck you can set deterministic uuids and stuff to avoid the getrandom calls
<Dezgeg> or yes this haveged thing but having custom code for that in the initrd is kind of annoying probably
<Dezgeg> or actually, if the hisi-rng module works for that hardware, then include that on the initrd
sphalerite has quit [Ping timeout: 252 seconds]
sphalerite has joined #nixos-aarch64
orivej has joined #nixos-aarch64
Acou_Bass has joined #nixos-aarch64
Acou_Bass has quit [Ping timeout: 252 seconds]
Acou_Bass has joined #nixos-aarch64
orivej has quit [Ping timeout: 252 seconds]
<gchristensen> my life is compiling kernels now
<gchristensen> Dezgeg: the hisi-rng seems to have improved it! now fixing no networking due to missing MLX modules.
<Dezgeg> ok, great
<gchristensen> do you not sleep much, Dezgeg?
<Dezgeg> depends... it's like afternoon here right now
<gchristensen> Dezgeg: maybe you have some more recommended kernel modules :) lspci seems a bit parse https://gist.github.com/grahamc/2b7a14dd8336d0a95be375a76a903c30
<gchristensen> sparse*
<gchristensen> [ 25.985314] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
<gchristensen> [ 26.097292] IPv6: ADDRCONF(NETDEV_UP): eth1: link is not ready
<gchristensen> [ 26.373837] IPv6: ADDRCONF(NETDEV_UP): bond0: link is not ready
<gchristensen> [ 26.541831] bond0: Enslaving eth0 as a backup interface with a down link
<gchristensen> [ 26.709818] bond0: Enslaving eth1 as a backup interface with a down link
<gchristensen> none of my nics have up links after boot, but do at ipxe time
<Dezgeg> all the modules should be there given you have the eth* devices showing up
Thra11 has joined #nixos-aarch64
orivej has joined #nixos-aarch64
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-aarch64
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-aarch64
orivej has quit [Ping timeout: 252 seconds]
jtojnar has quit [Remote host closed the connection]
Thra11 has quit [Ping timeout: 260 seconds]
LnL7 has joined #nixos-aarch64
LnL7 has quit [Remote host closed the connection]
LnL has quit [*.net *.split]
LnL has joined #nixos-aarch64
jtojnar has joined #nixos-aarch64