golinux | Unfortunately most of the infra team who could fix it are probably zzzzzzzing | 00:04 |
---|---|---|
Wasp | hey, can anybody give me a hint what I'm supposed with apt "The following packages cannot be authenticated". Almost every source, also recent sources, tell me to do `apt-key update` but apt-key tells is me depricated but gives no hint what to do instant... | 01:52 |
gnarface | Wasp: catch 22 | 01:54 |
gnarface | key expired | 01:54 |
gnarface | needs good key to get updated key package | 01:55 |
gnarface | there's a number of reasons this could have happend, but normally it should not happen after your first install | 01:55 |
Wasp | gnarface: sorry, what do you mean by catch 22? | 01:55 |
gnarface | hmm. are you familiar with something called a "chicken and egg problem?" | 01:55 |
Wasp | gnarface: nah, have it running for a while | 01:56 |
Wasp | so what is the "correct" way to solve it? | 01:56 |
gnarface | manually get the updated key package first | 01:56 |
gnarface | (or just feed the contents of the key to apt-key directly like it's asking) | 01:57 |
Wasp | I also checked apt-key list but there is no expired key | 01:57 |
gnarface | are you still on jessie? | 01:57 |
Wasp | guess so | 01:57 |
gnarface | there was an issue i remember about this | 01:57 |
Wasp | /etc/*version* says "9 ascii" | 01:57 |
gnarface | it had happend to me too, but it should have been resolved long ago... | 01:58 |
gnarface | what's the contents of your /etc/apt/sources.list file? | 01:58 |
gnarface | maybe you're using outdated urls | 01:58 |
gnarface | they changed auto.mirror.devuan.org to deb.devuan.org but there are still cached DNS entries for the old one floating around pointing to outdated repos | 01:58 |
Wasp | I don't know, was trying to check as well but no clue how (also cannot reach devuan.org) | 01:58 |
gnarface | oh, the website is down right now, that has been noticed this afternoon | 01:59 |
gnarface | unfortunately it's the middle of the night for everyone who can do something about that right now | 01:59 |
Wasp | ah okay :) Greetings from central europe too ;) | 01:59 |
gnarface | here, start with this, copy and paste the contents of your /etc/apt/sources.list file to paste.debian.net | 02:00 |
Wasp | I can confirm, yeah middle of the nighth here ;) | 02:00 |
gnarface | i can proofread your sources.list so at least we can be sure that one is correct | 02:00 |
gnarface | or if you're too paranoid for that, just make a backup of the file and put this in it's place: http://paste.debian.net/1072523/ | 02:01 |
gnarface | (make sure you're really on ascii first) | 02:01 |
Wasp | grrrr "Invalid format for name (no special chars, max 10 chars)" -- always a good idea to empty all input fields | 02:01 |
gnarface | yikes, i can't imagine why you'd have any special characters in there... | 02:02 |
KatolaZ | Wasp: wazzup? | 02:02 |
Wasp | and max 10 chars funny paste site | 02:02 |
KatolaZ | Wasp: will be here just for a couple of minutes... | 02:03 |
Wasp | http://paste.debian.net/1072524/ | 02:03 |
gnarface | looks right at first glance | 02:04 |
Wasp | KatolaZ: sorry don't know who you are but having issues to update packages and possible the whole system | 02:04 |
Wasp | because of "The following packages cannot be authenticated" | 02:04 |
KatolaZ | Wasp: can you please post the whole output you get? | 02:05 |
KatolaZ | (on a pastebin not here) | 02:05 |
Wasp | apt-key list doesn't show any expired keys | 02:05 |
Wasp | sure | 02:05 |
gnarface | Wasp: i had an issue a while back where the key from the package didn't actually make it to apt-key for some reason... i forget why exactly but i thought this was a problem that later got fixed. it's possible you could run into it still though if you had not updated since before that.... | 02:06 |
Wasp | KatolaZ: https://nopaste.xyz/?a97ba2e6625dc912#futX+bIqwYK6Wa3ehmbrYd97VIGrREG9Zn2Khf55C34= | 02:07 |
KatolaZ | Wasp: are you mixing repos? | 02:07 |
Wasp | what you mean? I have fish and vivaldi as an extra repo | 02:08 |
KatolaZ | are you sure you don't have what was called "debian-multimedia.org"? | 02:08 |
gnarface | hmmm.... | 02:08 |
gnarface | KatolaZ: (it's deb-multimedia.org now) | 02:08 |
KatolaZ | yep | 02:08 |
Wasp | yeah maybe i have | 02:08 |
fsmithred | apt-get install devuan-keyring | 02:09 |
KatolaZ | Wasp: that's the reason | 02:09 |
KatolaZ | nono | 02:09 |
KatolaZ | wait fsmithred | 02:09 |
Wasp | > cat /etc/apt/sources.list.d/deb-multimedia.list ⑂ 56d2770 | 02:09 |
Wasp | deb http://www.deb-multimedia.org stretch main non-free | 02:09 |
KatolaZ | please remove deb-multibedia | 02:09 |
KatolaZ | the repo has nothing to do with debian/devua | 02:09 |
KatolaZ | ~devuan | 02:09 |
KatolaZ | just remove it | 02:09 |
KatolaZ | then apt-get update | 02:09 |
KatolaZ | and apt-get upgrade | 02:09 |
KatolaZ | deb-multimedia is not needed since wheezy, at least | 02:09 |
KatolaZ | since all the packages that used to be there have been imported in Debian | 02:09 |
KatolaZ | by the debian-multimedia team | 02:10 |
KatolaZ | Wasp: ^^^ | 02:10 |
Wasp | okay, but even though i cannot remember way, but I'm sure I added it for a reason | 02:10 |
KatolaZ | Wasp: trust me, there is no reason | 02:10 |
se7en | I am unable to run apt-update nor fetch packages. Apt reports it is fetching from the IP 46.105.191.77 | 02:10 |
se7en | Is the server down now? | 02:10 |
se7en | It is the packages.devuan.org url | 02:11 |
KatolaZ | se7en: please use deb.devuan.org | 02:11 |
KatolaZ | we currently have issues on some of the domains | 02:11 |
KatolaZ | including packages.devuan.org | 02:11 |
se7en | Ok | 02:11 |
fsmithred | I would say it's working - I'm running live-sdk right now and it's installing packages | 02:11 |
Wasp | KatolaZ: I grant you that you know better maybet but at least I'm sure I've added it because I've (maybe just thought) needed something from there | 02:11 |
se7en | I have not heard of that url, is that a perment thing or will I have to change it back | 02:11 |
KatolaZ | Wasp: if it was libdvdcss, it's not needed any more | 02:12 |
KatolaZ | there is a replacement | 02:12 |
Wasp | KatolaZ: right! | 02:12 |
gnarface | Wasp: i have a guess as to why you added it. i note you've enabled non-free there, and for one of your commented out security updates lines in the sources.list, so probably you were looking for something like faac and made a mistake in your sources.list to confirm the erroneous assumption that it was only available in deb-multimedia | 02:12 |
KatolaZ | se7en: it has been a permament thing since at least one year ago | 02:12 |
Wasp | I remeber I needed libdvdcss because I couldn't find it in the regular repo | 02:12 |
gnarface | Wasp: yea, when you do that though, ONLY install libdvdcss from whatever repo, then immediately remove that repo | 02:13 |
gnarface | Wasp: (i think you should probably be getting libdvdcss2 from videolan.org's repo anyway) | 02:13 |
KatolaZ | Wasp: there is no need for it any more | 02:13 |
se7en | Ok, thank you KatolaZ | 02:14 |
se7en | Btw, has there been any new updates to setnet? | 02:14 |
KatolaZ | Wasp: https://wiki.debian.org/CDDVD#Video_DVD | 02:15 |
KatolaZ | you can use libdvdread4 instead | 02:15 |
gnarface | hmm. i had an issue where that didn't work in some cases, and i still needed libdvdcss2, but i can't remember the exact cause of the issue. | 02:16 |
KatolaZ | Wasp: alternatively, you can install libdvd-pkg | 02:16 |
KatolaZ | it will build libdvdcss2 for you from the repo | 02:16 |
KatolaZ | yeah gnarface you are right | 02:18 |
KatolaZ | some dvds are encrypted with CSS | 02:18 |
KatolaZ | those need libdvdscc2 | 02:18 |
KatolaZ | ~libdvdcss2 | 02:18 |
KatolaZ | the clean way is to install libdvd-pkg from contrib | 02:18 |
KatolaZ | it will download the tagged sources from upstream, build the deb package, and install it | 02:19 |
KatolaZ | I gotta go | 02:19 |
KatolaZ | bbl | 02:19 |
KatolaZ | o/ | 02:19 |
se7en | wait, KatolaZ | 02:20 |
se7en | my question on setnet | 02:20 |
KatolaZ | what? | 02:20 |
KatolaZ | se7en: which version are you using? | 02:20 |
KatolaZ | the latest available in the repos should be 0.4 IIRC | 02:21 |
* Wasp is up-to-date with apt-packages again | 02:21 | |
Wasp | KatolaZ: thank you | 02:22 |
KatolaZ | Wasp: yw | 02:22 |
Wasp | gnarface: thank you too ;) | 02:22 |
gnarface | no problem | 02:22 |
Wasp | what is the propper way to update the system/kernel? | 02:23 |
Wasp | refering to wikiepdia 4.9 is a little outdated | 02:23 |
se7en | 0.4, KatolaZ | 02:23 |
gnarface | Wasp: if you install the "linux-image-amd64" or "linux-image-i386" package (as appropriate for your computer's architecture) then you'll get the updated kernel packages along with everything else | 02:23 |
se7en | I am not using repo cersion | 02:24 |
se7en | Glad to see it's in there now | 02:24 |
KatolaZ | Wasp: you have 4.19 in ascii-backports | 02:24 |
KatolaZ | se7en: no new release | 02:24 |
KatolaZ | I have made some improvements but not released them yet | 02:24 |
KatolaZ | I hopw I will have time for that :) | 02:24 |
gnarface | Wasp: (same rule applies for linux-headers-amd64/linux-headers-i386) | 02:25 |
KatolaZ | gotta go, really | 02:25 |
KatolaZ | o/ | 02:25 |
Wasp | bye KatolaZ | 02:26 |
gnarface | Wasp: oh, sorry, overlooked the part about 4.9 being too old. yea, you'll want to grab the newer one from ascii-backports | 02:26 |
gnarface | if you're using nvidia drivers, you'll want to get them from there too along with it | 02:26 |
gnarface | or if you're using intel/amd graphics, you'll want to get mesa from there | 02:27 |
gnarface | certain generations of things work better when they're paired together | 02:27 |
Wasp | never managed to install nvidia here .. couldn't never get rid of f*** nouveau | 02:27 |
gnarface | oh, that's super easy but should have happened automatically if you didn't use the shell script installer from nvidia.com | 02:28 |
gnarface | the trick is to blacklist nouveau in /etc/modprobe.d | 02:28 |
Wasp | don't know anymore, gave up long time ago and just accepted that my machine freezes up randomly from time to time | 02:28 |
gnarface | well, the official drivers might not fix that, but chances are you ran into an easily preventable problem | 02:29 |
Wasp | don't have this problem with windows nor with freebsd though | 02:30 |
gnarface | you'd literally just have to have the official drivers successfully installed, and add "blacklist nouveau" to a file with the extension ".conf" in /etc/modprobe.d | 02:30 |
gnarface | bsd and windows both have commercial-level support contracts with NVidia | 02:30 |
gnarface | so they're not a fair comparison | 02:30 |
Wasp | what else should I copare it to though? plan9? ;) | 02:31 |
gnarface | heh | 02:31 |
gnarface | well the point is someone is paying NVidia to make it work for those two platforms, whereas they've been browbeaten to supporting Linux on their own dime via public outrage. a fact which is clearly represented in the quality of support | 02:32 |
Wasp | but supprised to hear that freebsd is paying nvidia | 02:33 |
Wasp | never heard before. Do you have any source for that? It's kinda supprising to me since freebsd is also a free OS | 02:34 |
gnarface | nvidia wouldn't support them at all, so i guess some people over there just purchased a license. it's something i heard from multiple people but i guess i don't have a link | 02:34 |
Wasp | I always just assumed somebody at nvidia likes freebsd | 02:35 |
Wasp | I mean they also provide drivers for linux | 02:35 |
gnarface | heh, no. those assholes hate everything open source stands for. | 02:35 |
gnarface | but they'll take any money that's green. | 02:35 |
gnarface | the demand for Linux support is partially justified to them by some high-end computing cluster hardware contracts | 02:35 |
Wasp | I get your point but -- at least at the beginning -- I would wonder that microsoft payed them to provide drivers | 02:36 |
gnarface | i wonder that too, but it's probably some percentage of global sales revenue | 02:36 |
gnarface | so it's probably a really shockingly high number | 02:37 |
gnarface | anyway, this is kinda OT stuff. we should rant about nvidia in #debianfork instead | 02:37 |
Wasp | haha | 02:37 |
Wasp | to be fair I like nvidia mostly ;) | 02:37 |
Wasp | there are two things I accept to be proprietary, and one is nvidia | 02:39 |
gnarface | is that one of those optimus dual-gpu laptops you've got there? | 02:44 |
Wasp | no laptop at all. Just a classic desktop | 02:44 |
gnarface | is the gpu a 600 series or later? | 02:45 |
Wasp | later | 02:45 |
gnarface | you should be able to get the nvidia drivers in the repo working | 02:45 |
Wasp | it's still in the 1000 generation .. i mean _still_, not again ;) | 02:45 |
Wasp | 8800 or 8600 don't recal atm | 02:46 |
gnarface | OH | 02:46 |
Wasp | have both | 02:46 |
gnarface | so that's actually older than the GTX 600 | 02:46 |
Wasp | I guess so ;) | 02:46 |
gnarface | 6** | 02:46 |
gnarface | ok nevermind then | 02:46 |
gnarface | the current official nvidia drivers already deprecated those cards anyway | 02:46 |
gnarface | you're stuck with nouveau | 02:46 |
gnarface | (or the official "legacy" drivers, which are pretty much useless now) | 02:47 |
Wasp | in summer they rip themself appart so have actually 3 of them and repair the other two from time to time | 02:47 |
Wasp | why is it useless? | 02:47 |
gnarface | not currently updated, so it won't work with anything still compatible with the rest of the distro | 02:48 |
gnarface | well, maybe glxgears or stuff like that | 02:48 |
gnarface | but not stuff that matters | 02:48 |
gnarface | not stuff nouveau can't do now anyway | 02:48 |
Wasp | what does matter? ;) | 02:48 |
gnarface | hardware opengl acceleration | 02:48 |
Wasp | nouveau can freeze the system | 02:48 |
Wasp | but that's nothing I like to see | 02:49 |
gnarface | yea, it can, but there might be a workaround | 02:49 |
Wasp | current work arround is moving from vivaldi to opera (again) since it's happening much less though | 02:49 |
Wasp | somehow related to youtube | 02:49 |
gnarface | hmmm, interesting | 02:49 |
Wasp | something* | 02:49 |
gnarface | if that's a reproducible bug, you should see if that's something they know about in #nouveau | 02:50 |
Wasp | I thought nouveau is dead .. or at least going to be dead. Cannot remember details but just read that they're ^(tm)^ planning to remove it from somewhere ^(tm)^ | 02:51 |
gnarface | i don't know about that, but their progress has definitely been halted for newer cards | 02:52 |
Wasp | sorry, just stupid hear-saying with not even recalling the propper context :( | 02:52 |
gnarface | and they'll tell you themselves not to hold your breath on it ever working right | 02:52 |
gnarface | they'll recommend AMD cards across the board | 02:52 |
Wasp | haha | 02:52 |
Wasp | yeah, but for "classical" reasons (nothing really substantial) I'm not a big fan of AMD graphic cards | 02:53 |
Wasp | and to be fair, for real I'm a freebsd guy and there it runs/ran fine | 02:54 |
* Wasp also really have to upudate it's freebsd | 02:54 | |
gnarface | the only thing important about AMD in this context is just that they're visibly working on improving | 02:54 |
Wasp | amd on freebsd was always just a pain ** t** a** | 02:54 |
gnarface | i've mostly only used openbsd, and mostly only headlessly so i don't have any reference point for comparison | 02:56 |
Wasp | Kato*laZ said I should add contrib for libdvd-pkg but acutally there is no libdvd-pkg in contrib but libdvdcss and libdvdread | 02:59 |
gnarface | hmm, pkginfo.devuan.org suggests you're wrong about that | 03:00 |
gnarface | make sure you've added contrib and non-free to the end of all those lines in your sources.list | 03:00 |
gnarface | it should go "main contrib non-free" | 03:00 |
gnarface | and make sure you run "apt-get update" after making the change, but before searching for the libdvd-pkg package | 03:01 |
Wasp | well, `apt-cache search libdvd | wc -l` says: 0 | 03:01 |
gnarface | right, but that depends on your apt cache, which depends on sources.list | 03:01 |
Wasp | forget my last line | 03:01 |
gnarface | which will be wrong until you make the change | 03:01 |
gnarface | AND run apt-get update | 03:01 |
Wasp | however there id libdvdnav, libdvdnav4, libdvdread and libdvdcss2 | 03:02 |
Wasp | id=is | 03:02 |
gnarface | those are in main except for libdvdcss2, which is in non-free | 03:03 |
gnarface | oh | 03:03 |
gnarface | i mean which would be in non-free on other repos but isn't in devuan repos i mean | 03:03 |
Wasp | however my point/question here is, "libdvd" no single package (while the oppisite was kinda implied by KatolaZ) or do I still miss something? | 03:04 |
gnarface | i assume the version of it that libdvd-pkg builds will flag itself as from non-free though | 03:04 |
gnarface | yes, you missed something | 03:04 |
gnarface | the part where that must be your mistake | 03:04 |
gnarface | here, look at this: https://pkginfo.devuan.org/cgi-bin/d1pkgweb-query?search=libdvd-pkg&release=any | 03:05 |
gnarface | shows as present in contrib for every release | 03:05 |
Wasp | http://paste.debian.net/1072530/ | 03:05 |
gnarface | ok, i'm not gonna lie, you're frustrating me because i already told you and clearly you skimmed it | 03:06 |
gnarface | but stand by and i'll give you an edited version of that | 03:06 |
Wasp | probably I overread it :/ | 03:06 |
Wasp | however realizing something | 03:07 |
gnarface | Wasp: like this http://paste.debian.net/1072532/ | 03:08 |
gnarface | Wasp: *THEN RUN apt-get update* | 03:08 |
Wasp | I know apt update | 03:08 |
Wasp | however added it to ascii-security what is not helpful | 03:08 |
Wasp | have it now | 03:09 |
gnarface | yea because it needs to be on the "ascii" and "ascii-updates" lines too | 03:09 |
gnarface | basically you want those parameters to be the same for every line in there | 03:09 |
gnarface | whatever you set them to | 03:09 |
gnarface | (it's ok to remove contrib and non-free when you're done with this, if you don't want those packages to update further without your knowledge - stuff in contrib and non-free tends to have lower quality than the rest of the stuff) | 03:10 |
Wasp | yeah, didn't know that (even though I added it to the wrong line by mistake, I always understood it as some kind of parameters to the repos) | 03:11 |
Stacker | Hello! I am having issues connecting towards devuan.org (and subdomains packages.devuan.org) anyone else experiencing this issue? I am not getting a ping on the addres that 'dig devuan.org' gives me. | 10:04 |
debdog | yah, it's down | 10:06 |
Centurion_Dan | Stacker: for mirrors use deb.devuan.org - the main website is down currently and we're trying to get things back online. | 10:07 |
Stacker | Centurion_Dan: Thanks for the info! | 10:08 |
Stacker | Alright! It's now fetching the package lists correctly, thanks alot! | 10:09 |
sokan | buZz: I've ordered my T420 \o/ | 11:47 |
buZz | w00t | 11:47 |
buZz | 420 erry day | 11:48 |
specing | sokan: so you are fine with the backdoors? | 11:48 |
sokan | well... the order will take place on tuesday but I'll have it on Thursday probably | 11:48 |
buZz | intel ME? you can flash that away on T420 | 11:48 |
specing | nope | 11:48 |
sokan | specing: yeah... beggars can'tbe choosers | 11:48 |
buZz | yep | 11:48 |
sokan | :\ | 11:48 |
specing | ME is hard silicon, can't flash that away | 11:48 |
buZz | you can | 11:48 |
buZz | you can neuter ME by replacing the code of it in bios | 11:49 |
sokan | specing: my desktop is amd ryzen 1700 so it also has some sort of backdoor | 11:49 |
buZz | it'll run without network capability etc :P | 11:49 |
sokan | we can't avoid it these days afaict | 11:49 |
buZz | well, we can | 11:49 |
buZz | you just need to pay more ;) | 11:49 |
sokan | haha | 11:49 |
sokan | mac? | 11:49 |
buZz | nearly all thinkpads are also made for NSA etc | 11:49 |
sokan | :P | 11:49 |
buZz | without ME on it | 11:49 |
sokan | :O | 11:49 |
buZz | wait, without AMT on it | 11:50 |
buZz | it'll still have ME | 11:50 |
sokan | hahah | 11:50 |
buZz | all cpus after p4 iirc need ME to function | 11:50 |
sokan | I've give up on trying to be 100% free frome everything :( | 11:50 |
specing | buZz: you cannot "replace" the code if it, it is digitally signed | 11:50 |
sokan | s/give/given | 11:50 |
buZz | specing: you can | 11:50 |
sokan | trying my best to just avoid corps, google etc at least :( | 11:50 |
buZz | https://hackaday.com/2016/11/28/neutralizing-intels-management-engine/ | 11:50 |
buZz | specifically https://hardenedlinux.github.io/firmware/2016/11/17/neutralize_ME_firmware_on_sandybridge_and_ivybridge.html | 11:51 |
specing | sokan: just get a power9 talos, it even has open source microcode | 11:51 |
sokan | buZz: chances of bricking the machine? (I haven't yet read the articles) | 11:51 |
sokan | specing: :O :O | 11:51 |
buZz | sokan: low | 11:51 |
buZz | you can flash ME back onto it | 11:51 |
buZz | the process has you make a backup copy even | 11:51 |
specing | buZz: "neutralizing" is very different to "flash that away". me_cleaner only reduced the firmware to the "bare minimum", whatever that means | 11:52 |
specing | reduces* | 11:52 |
buZz | exactly, so neutralizes the impact | 11:52 |
buZz | as much as humanly possible | 11:52 |
buZz | which is verrrry far from 'cant' | 11:52 |
sokan | none the less! I'll have a good quality laptop (afai've been told anyway) with devuan on it \o/ | 11:55 |
buZz | \o/ | 11:55 |
sokan | I'm planning on trying out disk encryption on it as well | 11:55 |
KatolaZ | Devuan web accessible already at dev-1.org and dev-one.org | 11:58 |
KatolaZ | (HTTP + HTTPS) | 11:58 |
KatolaZ | working to restore www.devuan.org | 11:58 |
buZz | nice nice | 12:02 |
buZz | KatolaZ: what happened? | 12:03 |
booyah | when your cat thinks server's main power cable is a toy | 12:35 |
sokan | lol | 12:38 |
sokan | oh buZz ! I forgot to comment on " 420 erry day ": 'tis a good number name :P | 14:02 |
buZz | :P lol | 14:03 |
sokan | my only worry is how to handle 120 gb ssd... 120 seems... | 14:05 |
sokan | well, let's say that it feels like a few MB :S | 14:05 |
specing | 120 is plenty | 14:10 |
smk | justice? | 18:17 |
smk | whyfor is devuan = [#devuan] Join us, and celebrate in freedom and justice | 18:26 |
Xenguy | smk: Source? | 18:28 |
smk | [12:16] -ChanServ- [#devuan] Join us, and celebrate in freedom and justice | 18:30 |
smk | its the welcome message as set by channel ops | 18:30 |
KatolaZ | smk: have never see it... | 18:31 |
KatolaZ | anyway, what bothers you? | 18:31 |
smk | i'm just curious about the meaning of the justice part | 18:33 |
smk | were the devs that forked with new project wronged by systemd camp in some way? | 18:34 |
buZz | sokan: keep in mind that small SSDs will perform worse then bigger ones | 19:39 |
buZz | marginally, but some | 19:39 |
sixwheeledbeast | Yer aim to keep them less than 60% full at anytime | 19:45 |
buZz | yeah something that gets harder to do with smaller SSDs aswell ;) | 19:45 |
Xenguy | sixwheeledbeast: Why aim to keep an SSD <60% full? | 20:08 |
KatolaZ | Xenguy: 'cause it will last longer | 20:09 |
Xenguy | OK, just curious, why will that cause it to last longer? | 20:09 |
Xenguy | I am using my first SSD in this laptop, and it is currently 80% full | 20:10 |
sixwheeledbeast | there is the wear side but also the more full they are the slower they get | 20:11 |
Xenguy | Huh, good to know (fortunately the speedup compared to using traditional HD tek probably more than makes up for that?) | 20:12 |
buZz | all SSDs have more flash then they report as data | 20:13 |
rwp | If it is a good SSD vendor then they will have overprovisioned it sufficiently that lifetime will not be a problem. | 20:13 |
buZz | if they would have -exact- 120GB for a 120GB model, it would die -quickly- beyond belief | 20:13 |
sixwheeledbeast | The drive has to read all partially filled blocks before writing to them whereas an empty block can be written to straight away | 20:13 |
Xenguy | df tells me this SSD has a 215Gb capacity | 20:14 |
KatolaZ | Xenguy: ssd are written by block | 20:14 |
rwp | I have never seen an Intel SSD nor a Samsung EVO fail due to NAND wearing out. But I have a stack of failed OCZ SSDs. | 20:14 |
specing | Xenguy: I don't think it should matter | 20:14 |
KatolaZ | and the disk moves the blocks around when writing | 20:14 |
specing | Xenguy: some people swear to leave a gigabyte or so free at the end | 20:15 |
rwp | Xenguy, df tells you how much file system free space is there. It cannot peer behind the curtain and see what the vendor is using for wear levelling. | 20:15 |
KatolaZ | hence if it has more free space, each cell will be used less frequently | 20:15 |
sixwheeledbeast | All SSD's have some over provisioning space wise. There is also TRIM and wear levelling which increases life | 20:15 |
KatolaZ | sixwheeledbeast: over-provisioning is often reserved for bad blocks | 20:15 |
* Xenguy decides to read up on SSD's ... | 20:15 | |
rwp | I question the practice of leaving space un-written to. However with use of TRIM that is possible. | 20:15 |
rwp | Remember that without TRIM the drive has no way of knowing that the file system free space is actually unused. It's just data. | 20:16 |
buZz | keep in mind that -reading- from SSDs will cause -writes- aswell | 20:16 |
buZz | as the nand cells will loose power from reading, and will need to get re-written after X time | 20:16 |
rwp | I have worked with a lot of SSDs and have not yet enabled TRIM on any of them. I am in the camp that says good vendors over-provision them and wear level in the background. | 20:17 |
rwp | TRIM should help a low quality vendor that has very little over-provisioning however. | 20:17 |
buZz | well for just typical 3-5 year usage you wont run into the issues quickly | 20:18 |
buZz | unless you have seriously misconfigured something :P like programs that write 100GB logfiles a day or something | 20:18 |
buZz | you dont need to enable trim btw, could just run fstrim every week or something | 20:19 |
rwp | I am many years into an Intel prototype SSD and I have worked it hard for all of these years. It's in my Thinkpad T61p. I moved it from my T43p into it. Could it be 10+ years old now? Time flies! | 20:20 |
buZz | hmhm, come back when you tested 100 of them next to each other ;) | 20:21 |
rwp | That is how I ended up with a stack of OCZ drives. Those dropped out rather quickly. It was a school environment. | 20:21 |
buZz | right, so you -have- actual data of SSDs failing but fail to see relevance to n=1 testsets next to it ;) | 20:22 |
sixwheeledbeast | Yes, over provisioning is for bad blocks and drive firmware but also it needs somewhere to move data to while shuffling write operations about. | 20:22 |
buZz | sixwheeledbeast: -and- read operations | 20:22 |
buZz | which cause writes | 20:22 |
sixwheeledbeast | exactly | 20:23 |
buZz | so a full SSD will also copy data slower | 20:23 |
sixwheeledbeast | :nod: | 20:23 |
rwp | (me laughs) I have a handful of Intels and a couple of dozen of Samsungs running. No problems with either of those so far. | 20:23 |
sixwheeledbeast | I still use hard disks for the majority of my system. | 20:24 |
buZz | rwp: no offense, but keep it around for such occassions ; https://i.imgur.com/pJPYN6p.png | 20:24 |
buZz | :D | 20:24 |
sixwheeledbeast | :) | 20:24 |
rwp | buZz, Without using TRIM the drive firmware doesn't know the drive isn't full. But a drive should be designed to operate with the over-provisioned blocks sufficiently without requiring TRIM. | 20:25 |
buZz | eh, no | 20:25 |
buZz | drive firmware always knows | 20:25 |
* rwp laughs :-) | 20:25 | |
rwp | buZz, regarding that the drive firmware always knows, no, that is incorrect! :-) Which I only say due to your funny that you shared. | 20:26 |
buZz | rwp: ok, offer evidence :D | 20:26 |
rwp | It can only know if you use trim. And cannot if you do not. | 20:26 |
buZz | trim is 'mark empty blocks as fully empty' | 20:26 |
buZz | it cant mark those without KNOWING they are empty previous | 20:26 |
rwp | Basically, yes. | 20:26 |
buZz | and doesnt have to be done by OS | 20:26 |
rwp | Right. | 20:26 |
buZz | so how does it know, magic? | 20:27 |
rwp | Negative! *Must* be done by the OS. Otherwise it does not know. | 20:27 |
buZz | ah yeah, they have nontrimming internally like you mention with drawbacks | 20:28 |
buZz | > By 2014, many SSDs had internal background garbage collection mechanisms that worked independently of trimming. Although this successfully maintained their performance even under operating systems that did not support trim, it had the associated drawbacks of increased write amplification and wear of the flash cells.[5] | 20:28 |
buZz | which knows the cells are empty , etc | 20:28 |
rwp | By 2014? That is probably talking about some of the newer entries to the market. | 20:29 |
buZz | well thats like 2 computer upgrades ago | 20:29 |
rwp | I guess I will say that I used to work in the cpu design lab but when that was shut down the people all scattered to other places. Among them various SSD vendors. I still buddy around with them and we will often talk about SSDs and how they work. I have had these conversations often with the people who are doing the work. | 20:30 |
rwp | If you want to know how much NAND is actually on the device then look at the board and count the chips. It can be worked out that way. Then look at the advertised size. The difference is their over-proisioning. | 20:31 |
buZz | lol | 20:32 |
rwp | A file system might write a 4k block (used to be 512 bytes) but NAND flash is written in blocks of megabytes. Therefore there is always backing cache ram to store and write data. | 20:32 |
specing | rwp: which cpu design lab? | 20:32 |
rwp | PA-RISC | 20:32 |
buZz | that makes you over 60 years old? | 20:33 |
rwp | Then Itanium. But by then I was working in the tools lab. | 20:33 |
rwp | buZz is pretty good with the math! :-) Not quite that old yet but getting there very quickly. | 20:33 |
buZz | 59 ;) | 20:33 |
sixwheeledbeast | 2 upgrades? I build a machine about every 8 years... | 20:33 |
buZz | we dont all live in prosperous nations ;) | 20:34 |
rwp | It's OT but I can't resist. "I'm not old." https://www.youtube.com/watch?v=t2c-X8HiBng | 20:34 |
sixwheeledbeast | er? | 20:35 |
Nephiel | Hi there. Was the devuan.org server outage fixed? | 20:49 |
Nephiel | I seem unable to run apt-get update on some servers due to a hash sum mismatch | 20:50 |
xinomilo | mirrors could be syncing, try again in a short while. | 20:51 |
xinomilo | outage seems fixed : https://dev1galaxy.org/viewtopic.php?id=2715 | 20:51 |
Nephiel | "Could not connect to packages.devuan.org:80 (2001:41d0:8:2c55::a2). - connect (101: Network is unreachable) [IP: 2001:41d0:8:2c55::a2 80]" | 20:53 |
Nephiel | *shrug* | 20:54 |
Jjp137 | use deb.devuan.org | 20:55 |
Nephiel | "W: Failed to fetch http://deb.devuan.org/merged/dists/ascii/main/Contents-amd64.gz: Hash Sum mismatch" | 21:00 |
Nephiel | Nothing to do but wait, I guess | 21:01 |
stripe | hi all, anyone know if the raspberry pi 3 image will work on the 3b+ same chip just clocked a bit faster, just want to build a small headless vpn client, so dont need wifi, cheers | 21:18 |
xinomilo | stripe is gone, but yes :) | 21:26 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!