<Ashy>
i wonder if nix could be made to spit out binaries in that format
<colemickens>
I can't say that I feel that way about flakes actually. I thought I did at first, but it was largely just me getting used to restrictions and maybe some bugs.
<colemickens>
I'm pretty sure my problem is HW or ID-ten-T related
cjpbirkbeck has joined #nixos-chat
<samueldr>
Ashy: nix... yes, but not really, nix can do anything you ask it to do :)
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-chat
<infinisil>
Arghh, my dark theme for github has a bug, or github changed something that it doesn't fully work anymore
<infinisil>
And it doesn't just not work
<samueldr>
it ultra doesn't work&
<samueldr>
?*
<infinisil>
It actually works, but only after the bright page was shown for a split second
<infinisil>
So pretty much for every page click I get a flashing image of bright white..
<pie_>
whoda thunk being able to do stuff over the network is useful
<pie_>
took me a bit to get pest the pebkac that i wasnt looking at the x11vnc error right in front of me and then i figured out you can disable the shmem feature
<pie_>
why write blog posts when you can write github issues
<bqv>
why write github issues when you can post on irc
<pie_>
why not both
<colemickens>
well, I found the source of some of my issues today, from the zpool status, I quote, "too many errors"
<sphalerite>
colemickens: that can also be due to bad ram :p
<colemickens>
-_-
<sphalerite>
do you have ECC RAM?
<sphalerite>
and do you have any redundancy on the pool?
<sphalerite>
and which type of errors are you getting?
<colemickens>
so, I also have my HD being mounted as a hyperV VM disk drive
<colemickens>
and windows has been crashing, so that also is going on
<sphalerite>
If they're all checksum errors and they're appearing across all the devices it's pointing suggestively towards the RAM again ;)
<colemickens>
no ECC, no redundancy in pool (the only valuable thing is my unpushed partial commit and my current tabs).
<colemickens>
the thing is, the files that are corrupt are stuff like Firefox's DBs that live in memory for long times probably
<colemickens>
and those files would never be open when its running as a VM. :/
<sphalerite>
though who knows what kind of fun things happen if the kernel (hypervisor or guest) is in bad RAM :D
<sphalerite>
how did the overnight memory test look?
<colemickens>
:| yeah. time to call it a day and start a better test.
<pie_>
since i spent like 3 days on it i ahve to show everybody :P
<pie_>
this was a sidetrack while i started waiting for a kernel compile at some point...time to get back to that
<pie_>
or think of a way to make a bring-up script for this (probably involves the nix stuff sshing into termux)
<pie_>
hmm, can an app launch another app?
<bqv>
how ...terrifying :p
<bqv>
i've never been keen on the goddamn tiny displays of phones, bu that is pretty cool tbh
<bqv>
pie_: gg :D
<pie_>
i hate phones
<pie_>
but i seem to be using mine a lot recently
<bqv>
i think they can be mitigated for, ish
<pie_>
it doesnt help when the keyboard is covreing half the screen of the vnc client
<pie_>
soooo tiny screen?
<pie_>
how about even tinier screen
<bqv>
yeah, exactly
<pie_>
should i submit this to the newsletter
<cole-h>
I'm getting SERVFAILs randomly
<cole-h>
I guess Orbi routers just suck
FRidh has joined #nixos-chat
<viric>
Sometimes I'm surprised for seeing files with owner 666 in my fs
<viric>
until I recall that I often run chown when I mean chmod
<sphalerite>
hahaha
<bqv>
lol
<sphalerite>
do you use chmod 666 that much? D:
<viric>
I bet many religions have a weaker foundation than this
<viric>
sphalerite: maybe two or three times a year, I admit
<viric>
(unix sockets things, most of the time)
xd1le has quit [Read error: Connection reset by peer]
xd1le has joined #nixos-chat
cjpbirkbeck has quit [Quit: Goodbye, take care]
cole-h has quit [Quit: Goodbye]
sorear has quit [Read error: Connection reset by peer]
jared-w has quit [Ping timeout: 260 seconds]
savanni has quit [Ping timeout: 272 seconds]
sorear has joined #nixos-chat
<sphalerite>
anyone have HDD buying tips? I want high-capacity (8TB+) drives but non-SMR, and I heard the manufacturers aren't too open about whether their drives are SMR or not.
jared-w has joined #nixos-chat
<sphalerite>
I'm considering the Toshiba MG07ACA12TE
Guest88372 has joined #nixos-chat
<eyJhb>
sphalerite: they have started bashing them for it
<eyJhb>
So I don't think it is a issue anymore
<eyJhb>
But isn't the ironwolf from seagate the goto ?
<sphalerite>
the toshiba one is a bit cheaper :p
<sphalerite>
eyJhb: well, people have been bashing Donald Trump for a while…
<eyJhb>
But it is, Toshiba, soooo
<eyJhb>
I would check some benchmarks
<eyJhb>
Yeah but I think WD realeased a statement for it
<eyJhb>
They were the biggest offenders AFAIK
<sphalerite>
I don't care much about performance, I just want it to store lots of data. Anything above 80MB/s write is enough for me
<lejonet>
sphalerite: in general, the 8TB+ disks use either CMR or explicitly state SMR (because they are archival disks, i.e. WD purple), I have personally good experience with WD Red/Red Pro (6TB) and Toshiba N300 (6TB, tho I do believe those are EoL), I have 2 10TB running in one of my ceph nodes, but I can't remember the make and model
<sphalerite>
lejonet: alright, thanks :)
<lejonet>
sphalerite: I can double check the 10TB when I get home :P
<sphalerite>
I think I'll just buy two of the toshibas
<eyJhb>
I usually go WD, but I might have a fetish with them sphalerite
<eyJhb>
Just like Intel
parsley936 has joined #nixos-chat
<lejonet>
eyJhb: WD is generally my goto too, but I ensure that I always use at least 2 different manufacturers in my ceph nodes
<lejonet>
especially now as the WD Golds costs roughly as much as WD Reds, so no point in not going for WD golds
<lejonet>
or w/e they've renamed them to now, iirc they announced that they have or are going to rename them
<etu>
Avoid WD Red nowadays, get at least WD Red+
<etu>
WD Red is WD Red- :p
<lejonet>
etu: you're thinking about the SMR debacle?
<etu>
lejonet: yes
<lejonet>
Because that is "just" the ones in 2-6TB
<lejonet>
and if the goal is "only" 80MBps+ write, the SMR disks would be viable still
<lejonet>
(Unless you're going to have 24/7 writes, then they'll fall behind hard)
<lejonet>
If other enterprise disks are on the table, the seagate exoz isn't bad either
<etu>
lejonet: Depends if you use ZFS or not
<lejonet>
etu: that is true
<lejonet>
Do you have any links to observations or tests that show that ZFS interacts with SMR more badly than for example ext4? (asking out of actual curiosity, not trying to refute your claim)
<lejonet>
etu: ah yeah, the biggest achilles heel that SMR have are many, small writes, especially random ones
<lejonet>
Would've been interesting to be a fly on the wall during the meeting after the whole SMR debacle hit the wall, when they try to track down how they thought this was a good idea to sneak in :P
<adisbladis>
lejonet: I'm guessing the thought process was something like this: "How can we make these cheaper? You know all those SMR platters we make? Those are cheaper."
<lejonet>
adisbladis: SMR is pure firmware tho
<lejonet>
adisbladis: but yeah, probably something along those lines "How can we push these platters to higher capacity without actually adding more platters?"
<infinisil>
Very easy to confuse this with the standard ping dbinary
<elvishjerricco>
Does `boot.cleanTmpDir` slow down boot noticably? I imagine recursively deleting a directory with potentially several gigabytes stored amongst many files could take a little while.
<elvishjerricco>
Wondering if doing a zfs rollback of /tmp to an init snapshot would be better, and what systemd ordering to use for such a thing. Maybe After=zfs-import Before=zfs-mount?
<eyJhb>
sphalerite: might be the way
<eyJhb>
It just seems easier, to do it like that, when I have to pull in multiple repositories
<colemickens>
(please note I have not kept up with any of the docs discussions and any opinions I have held have changed a few times so...)
<samueldr>
bqv: go read the RFCs
<bqv>
Theyre longer than the manuals
<samueldr>
yeah
<samueldr>
if you were only on IRC, you're only having a quarter of the discussion, if not less
<samueldr>
eyJhb: the current implementation is still docbook
<bqv>
That's why I rely on people to parrot the important stuff :p
<samueldr>
bqv: or cherry-pick what they want :)
<eyJhb>
Still more informed than Trump.
<bqv>
Lmao
<cole-h>
Nice, Twitter is ded
<cole-h>
Oh, guess not
<cole-h>
Guess it's just broken for me
<samueldr>
so, the TLDR of the docs status is that the RFC for choosing a doc format closed inconclusively with the conclusion that an RFC that proposed a specific format had to be opened
<samueldr>
building on the knowledge of the RFC for docs choosing a new RFC is now open selecting *a format only*
<samueldr>
well, proposing markdown, and is currently in FCP
<pie_>
i never got around to doing the thing last time
<bqv>
why thank you, bot
<taktoa[c]>
I think clever went to bed
<bqv>
intelligence never sleeps
<taktoa[c]>
lol
<bqv>
hooray, home manager is now a flake!
Jackneilll has quit [Ping timeout: 258 seconds]
Jackneilll has joined #nixos-chat
__Sander__ has quit [Quit: Konversation terminated!]
<elvishjerricco>
Playing with zrepl now. There's a lot of documentation, but there is no documentation on the `status` subcommand -_- So now I have to read a bunch of Go code. And dear lord reading Go is awful.
xd1le has quit [Read error: Connection reset by peer]
xd1le has joined #nixos-chat
julm has quit [Quit: reboot]
julm has joined #nixos-chat
<bqv>
how bizarre.. nixpkgs-wayland seems to break my eval
<colemickens>
I blame anyone else.
<bqv>
colemickens: i can, with nixpkgs#nixUnstable, reliably crash nix, and the file it stops eval on is your flake
<samueldr>
colemickens: sounds good
<colemickens>
cole-h: (I think I got distracted and forgot to come back)... I bet you could totally modify flake.nix to have the 'beta' attribute btw. I could give you pointers for it if you wanted. Otherwise if you open an issue I'll do it sometime in September.
<energizer>
is there a program to change pitch of audio playback?
<worldofpeace>
bqv: oooh yay!
<bqv>
i mean it's undeniably a nix bug, but it always stops exactly at that pkgs/obs-studio/default.nix file, and has a really spooky effect on my pc (i think it actually puts all my cores at 100% or something, which is fucking impressive)
<bqv>
worldofpeace: :D
<worldofpeace>
energizer: yes, it's called a pitch shifter
<bqv>
audacity can probably do that
<energizer>
can i do it to live audio?
<bqv>
with ffmpeg or something, probably
<worldofpeace>
energizer: pulseeffects will work on the output sinks with what I think is the ladspa rubberband pitch shifter