rrq | right. thanks. I didn't scroll back very far... | 00:00 |
---|---|---|
CheesyPastries | Definitely using the FreeBSD one and it obviously doesn't have the Devuan patches (which is silly since according to that guide systemd doesn't work with FreeBSD's compatibility so Devuan is probably the OS of choice for this) | 00:01 |
CheesyPastries | If it had the patches it should've been aware of the different versions | 00:01 |
rrq | this: https://git.devuan.org/devuan-packages/debootstrap | 00:01 |
CheesyPastries | Is it possible to install the Devuan version on FreeBSD? I don't know if the framework would be in place to get it | 00:02 |
gnarface | there's the git link. try to build it | 00:02 |
CheesyPastries | I'm reading it now | 00:02 |
gnarface | i really don't think they had to change very much but i don't know for sure | 00:03 |
rrq | the HEAD of the suites/ascii branch is the ascii version | 00:04 |
CheesyPastries | It looks like i might be able to even run it without building it. Apparently it can run directly off the source tree according to the readme on the page | 00:04 |
gnarface | yea it might just be scripts with little or no binaries... i seem to recall that being the case... | 00:04 |
CheesyPastries | I'll give reinstalling with this version a shot then | 00:06 |
se7en | f | 00:20 |
CheesyPastries | I did a full reinstall with the Devuan debootstrap but the error persists | 00:52 |
gnarface | CheesyPastries: paste.debian.net with the error | 00:52 |
CheesyPastries | It complains about invalid signatures even after getting devuan-keyring and devuan-archive-keyrings added | 00:52 |
gnarface | you also need the debian ones probably | 00:52 |
gnarface | CheesyPastries: i still have no new ideas about this but i want to see the exact error just as a sanity check. you can paste it to me directly in /msg or you can use paste.debian.net | 00:54 |
CheesyPastries | https://paste.debian.net/1108448/ | 00:56 |
CheesyPastries | Note, I'm using a really f-ed up terminal and can't copy out of it so I had to retype that line by line. It should be accurate as I'm pretty good at that sort of thing | 00:56 |
gnarface | CheesyPastries: well... the thing about the repo being unsigned is weird. i'd expect maybe a error about the key being invalid... | 00:58 |
gnarface | CheesyPastries: there was an old bug at one point where you had to actually add the gpg key from the package to root's gpg keyring manually, because it wasn't checking in the right directory for it, but this isn't quite the error i'd expect from that glitch | 00:59 |
gnarface | CheesyPastries: still, it might be worth trying anyway as a test | 00:59 |
CheesyPastries | Agreed, I'm going afk to eat for a bit. I'll be back in about an hour. Post any thoughts you have and I'll try them when I get back and update you | 00:59 |
CheesyPastries | The help is much appreciated BTW | 01:00 |
CheesyPastries | I'm back now | 01:52 |
CheesyPastries | How would I go about adding the repository key to the root keyring? | 01:53 |
gnarface | apt-key add [filename] | 02:03 |
gnarface | as root | 02:03 |
gnarface | CheesyPastries: ^ | 02:03 |
CheesyPastries | Ah, I've actually already done that. and I just did an apt-key list and the keys are included | 02:03 |
gnarface | make sure they're the right ones | 02:04 |
gnarface | is there any chance there could be some sort of proxy issue? | 02:06 |
gnarface | http deep packet inspection? | 02:06 |
gnarface | anything | 02:06 |
gnarface | ? | 02:06 |
gnarface | i'm really scraping the bottom of the barrel for ideas here | 02:07 |
CheesyPastries | I'll double check they are correct | 02:07 |
gnarface | the only other thought i have is to try to make sure the signature file is actually on the repo, get it with a web browser, and check it manually with gpg | 02:07 |
gnarface | but i don't have the commands for that off the top of my head, you'd have to check the gpg manpage | 02:07 |
CheesyPastries | The only sort of proxy issue would be how the jail connects to the internet through the bridge connection on the device | 02:07 |
gnarface | yea the one thought i had was just that maybe it thinks the repo isn't signed because it's not even hitting the repo, but it's getting a valid http response from something else so you get this red herring error maybe | 02:08 |
gnarface | i have a router that when upstream connection fails, it redirects all DNS requests to itself | 02:09 |
gnarface | and then it responds with a valid http response page containing some error, but it's not a 404 page | 02:10 |
gnarface | so some stuff (like slashdot's javascript) fetches that page and then chokes hard on the response, having no logic to distinguish a valid http response from the wrong server as anything untoward | 02:11 |
CheesyPastries | But how did I install devuan-keyring from the repo if I'm not hitting the repo? | 02:12 |
gnarface | (to add to the fun, the page on the router redirects to itself, adds some junk to the query string that slashdot's javascript just blindly copies over, and then that puts the page into a redirect loop) | 02:13 |
gnarface | hmm | 02:13 |
gnarface | that's a good question though | 02:13 |
gnarface | i wouldnt' expect you to get a valid devuan-keyring package from your own router | 02:13 |
CheesyPastries | agreed | 02:13 |
gnarface | i would expect it from a MITM attack though maybe... | 02:13 |
gnarface | you double-checked the IP addresses that got resolved for you, right? | 02:14 |
gnarface | i mean, we're really running dry on ideas here | 02:14 |
CheesyPastries | Could it be related to me setting the jail to connect through the bridge0 instead of directly on an ethernet port? The rest of my jails connect through it just with the addition of VNET | 02:15 |
gnarface | i mean, as far as i know, only if it's mangling http requests somehow | 02:16 |
gnarface | though there is precedent for that too | 02:16 |
fsmithred | aptitude search ~i -F"%p# %v# %t#" | 02:17 |
gnarface | something to do with how amprolla serves unaltered debian packages via http redirect | 02:17 |
gnarface | it has been a sticking point before in some corner cases | 02:17 |
fsmithred | will show which suites the packages are from | 02:17 |
gnarface | also there was at least one or two incidents of some isp mangling https requests | 02:17 |
CheesyPastries | aptitude isn't installed | 02:18 |
fsmithred | I don't know an equivalent for apt-get or apt-cache | 02:19 |
gnarface | (or at least suspected of mangling https requests... in one situation it turned out to be the aforementioned 3rd party repo issue, but in the other incident i don't recall what was discovered) | 02:19 |
CheesyPastries | How could I test if it's mangling requests? | 02:24 |
CheesyPastries | fsmithred I force installed aptitude and the results of that command was mostly blackspace | 02:30 |
gnarface | there might be an error log somewhere | 02:30 |
gnarface | or try adding --verbose | 02:30 |
gnarface | or something like that | 02:30 |
gnarface | (check the man pages) | 02:31 |
gnarface | sometimes it's -v or -V | 02:31 |
gnarface | of course you could always just sniff the network traffic | 02:31 |
gnarface | like with wireshark or tcpdump | 02:31 |
gnarface | you'd have to have an idea of what types of http responses it's expecting though | 02:32 |
fsmithred | what was the part that wasn't blackspace? | 02:32 |
gnarface | my assumption would be that if it's an accident, the response will be WILDLY wrong, like containing the entirely wrong html page from an obviously wrong server. rather than a cleverly designed forgery | 02:32 |
CheesyPastries | fsmithred what seemed like a couple package names and stable separated out by a load of space | 02:34 |
fsmithred | I don't know what to make of that | 02:35 |
CheesyPastries | gnarface It'd be so weird if it was an actual MITM. This is a non-enterprise server so it seems an unlikely target | 02:35 |
fsmithred | is it possible for you to install devuan debootstrap and use that? | 02:35 |
CheesyPastries | Debootstrap was what I made this with. My guide cautions against using it for second-stage | 02:39 |
CheesyPastries | Also even trying to run it for the second-stage it gives back an invalid argument error and complains about / being mounted with noexec or nodev | 02:39 |
fsmithred | I've only used it to get the base system then chroot for everything else | 02:44 |
gnarface | CheesyPastries: well, it's worth noting, i have never tried this on bsd. i didn't think the different kernel would be a factor in this, but it's one of the few assumptions left unquestioned... | 02:44 |
fsmithred | but devuan forked debootstrap to add devuan-specific script(s) | 02:44 |
CheesyPastries | Yeah, this shit is weird | 02:45 |
gnarface | CheesyPastries: wait, you did try the devuan debootstrap already, right? | 02:46 |
CheesyPastries | fsmithred Last time I made it, I used the Devuan fork | 02:46 |
fsmithred | oh, ok | 02:46 |
gnarface | hmmmm | 02:47 |
gnarface | ok just thinking here, if it was a kernel issue, could it be a device issue? like permissions for the jail to read some hardware rng? | 02:47 |
gnarface | could gpg need hardware access on bsd? | 02:47 |
gnarface | i really don't know, i'm grasping at straws | 02:48 |
CheesyPastries | Want me to add another layer? It's not even pure freebsd, it's FreeNAS | 02:48 |
gnarface | oy vey | 02:48 |
CheesyPastries | I'm thinking of just giving up at this point | 02:48 |
gnarface | does it have qemu-kvm? | 02:49 |
gnarface | or something like that, which can put a linux kernel in a process? | 02:49 |
EHeM | FreeBSD has bhyve which appears to be sort of similar to Qemu. | 02:49 |
CheesyPastries | It already came with the linux compatibility layer and I think a preinstalled linux kernal | 02:50 |
CheesyPastries | Bhyve is for full VMs I believe | 02:50 |
gnarface | i would say, try that | 02:50 |
gnarface | well | 02:50 |
CheesyPastries | I could always fall back to that but I felt that a jail would be easier to manage the performance of (I don't have to assign a number of cores/RAM) | 02:51 |
EHeM | CheesyPastries: So is Qemu (though it can also emulate single processes). | 02:51 |
gnarface | first i would ask google if anyone else had ever had this problem, THEN i would try Bhyve | 02:51 |
gnarface | or Qemu if available | 02:51 |
gnarface | CheesyPastries: i'm with you on the performance thing, i'm just suggesting this mostly as a way to debug by logical deduction. if Bhyve were to fail the same way, that would seem to rule out the kernel | 02:52 |
CheesyPastries | I might do some extra googling and see if I can turn up anything. | 02:54 |
CheesyPastries | Qemu being able to emulate single processes might be interesting, then I could jail it with the programs that it needs instead of a full VM | 02:55 |
CheesyPastries | All this is so that I can get modern .NET since it hasn't been ported to FreeBSD yet and some apps I want to use are built around it instead of Mono now | 02:56 |
armin | ok so i'm reading a lot of markdown. what's a recommended way of reading these files? i keep using vim because that's what i'm used to, but i'm sure there's a better way, or? | 02:58 |
djph | nothing wrong with vim | 03:03 |
gnarface | armin: did you try "apt-cache search markdown" ? | 04:03 |
armin | gnarface: hello and thanks so much for the reply. i actually didn't. it wasn't really a technical question. i just keep wondering if i'm totally lost when i'm in my vim reading .md files - i'm used to doing that and i know what i read but at some point i was questioning what i do. | 04:16 |
armin | i should never have asked this. the one channel told me not to cross-post (sorry if that's annoying to someone, i just see 1->many vs. 1->1 here) | 04:18 |
gnarface | armin: ah. well i don't really know much about markdown or vim. all i can really do to help is point out that apt-cache supports regexp searches (grep pattern syntax, iirc) | 04:20 |
gnarface | i'm sure there are alternatives in there | 04:21 |
gnarface | there is a debian thing called popcon | 04:21 |
gnarface | they post the output somewhere... you might want to find that | 04:21 |
gnarface | (it's basically a list of packages by popularity) | 04:21 |
gnarface | (popularity being defined as number of times people who opted into popcon installed it) | 04:22 |
Jjp137 | here's Devuan's, for reference: https://popcon.devuan.org/ | 04:23 |
gnarface | ah, thanks Jjp137, i wasn't sure there was a stand-in for that yet | 04:23 |
Jjp137 | np; yeah I wonder how one would find it starting from the home page... | 04:23 |
Jjp137 | oh it's linked from pkginfo so there's that. | 04:24 |
agris | what do the packages firmware-raspi and raspi3-firmware do in Devuan ASCII? | 04:26 |
agris | and why would they not be installed by default in a raspiberry pi image? | 04:26 |
gnarface | agris: not sure but i think i recall raspbian doesn't even package the firmware??? | 04:28 |
gnarface | they just spray it on your drive with a script | 04:28 |
gnarface | iirc | 04:28 |
gnarface | so it might just be the same firwmare, or a subset of that, but packaged like a sane person would do | 04:28 |
agris | well I have a raspberry pi 3 here with Devian unstalled | 04:28 |
agris | Devuan | 04:28 |
agris | what would happen if I installed firmware-raspi ? | 04:29 |
agris | Would it keep blobs in /boot up-to-date for me in my system? | 04:29 |
gnarface | i don't know. they might also be just dev packages | 04:29 |
gnarface | you might want to ask about this in #devuan-arm | 04:29 |
gnarface | agris: for what it's worth, here on ceres, it says raspi3-firmware is a transitional package... | 04:32 |
agris | very weird | 04:33 |
Jjp137 | apparently it comes from Debian directly, so let's look at Debian's website for this one: https://packages.debian.org/stretch/raspi3-firmware | 04:33 |
Jjp137 | I guess you can look at the file list to get an idea | 04:34 |
agris | is reason the date on it is 2016? | 04:34 |
Jjp137 | but yea in sid/ceres it just becomes a transitional package | 04:34 |
agris | has the firmware really not seen any updates? | 04:34 |
Jjp137 | I'm guessing it's just Debian policy as usual to not update things unless necessary | 04:35 |
Jjp137 | but who knows | 04:35 |
agris | i'm very confused perhaps I can just tell you what I'm trying to do | 04:36 |
agris | I'm only using the pi3 with ethernet and a usb uart, running headless most of the time but HDMI is useful for debug | 04:36 |
agris | the system is having sporadic voltage issues even with official power supply | 04:37 |
agris | I want to underclock the pi to 600MHz, disable the bluetooth and wifi functionality of the soc, enable the internal hardware watchdog, disable turboing, and put the GPU into the lowest performance state possible | 04:38 |
agris | currently the CPU speed is variable between 600MHz and 1200MHz | 04:38 |
agris | I think when the cpu kicks up is when the bus voltage drops | 04:38 |
agris | I also do not understand this option | 04:39 |
agris | ## maximum amps on usb ports | 04:39 |
agris | max_usb_current=1 | 04:39 |
agris | I was not aware the soc had any control over USB current. I thought the USB power lines were just a straight trace from the microusb power input to the female sockets | 04:40 |
agris | I also do not understand the effect of this parameter off or on | 04:40 |
agris | I doubt USB UART is going to pull a full 5 amps so I don't think I need this option on. | 04:41 |
furrywolf | heh, my usb rs232 adapters pull a few ma. | 04:41 |
furrywolf | Some_Other_Guy: he seems to be perfectly comprehending the answers he gets. | 04:42 |
furrywolf | grr, wrong window | 04:42 |
agris | Devuan also does not seem to include the vcgencmd command which makes debugging this very hard | 04:43 |
agris | I don't know what provides vcgencmd | 04:43 |
agris | >On the Raspberry Pi 4, the hardware codecs for MPEG2 or VC1 are permanently disabled and cannot be enabled even with a licence key; | 04:45 |
agris | Pfft. fucking broadcom. It does not make sense to me how purposefully gimping their hardware can be a profitable business decision or why this decision was made. thankfully I'm not using the pi4 or the pi3 for video | 04:46 |
furrywolf | flir sells three models of each product line. each has the exact same hardware, exact same sensor, etc. the base model has the image scaled to half the size and lots of noise added. the middle model has the image scaled to 2/3rds the size and some noise added. the high-end model does not downscale the image nor add noise. so... broadcom is mild by comparrison? | 04:50 |
agris | furrywolf, ok, so you have hardware that is capable of a clear and unscaled image. You don't need to do anything to make improve the quality, no investment needs to be made. So why not sell un-gimped hardware at the lowest price? | 04:52 |
furrywolf | why sell things for a lower price when you can charge a higher price? | 04:53 |
furrywolf | i.e. capitalism. | 04:53 |
agris | It's not like it's costing the manufacturer any more money to sell the un-gimped version at the same price. in fact, it costs more to gimp the product. as an engineer has to go in there and implement the code to inject the noise and scaling | 04:54 |
agris | furrywolf, your response does not make any sense. | 04:54 |
furrywolf | sure it does. their goal is to make money, not to help you. | 04:55 |
agris | how is this 'capitalism''s fault? A capitalist would sell the better product at the same price because it doesn't cost him any more and it competes with others better | 04:55 |
djph | hell, intel (amd, etc) do it with their chips... | 04:55 |
furrywolf | no, that's not how capitalism works. | 04:55 |
furrywolf | prices rise until sales fall. | 04:56 |
onefang | Take that chat to #debianfork, this channel is for Devuan support. | 04:56 |
agris | Doesn't going out of their way COST broadcom/intel/amd etc MORE money than not gimping? | 04:56 |
agris | *Doesn't going out of their way to gimp the product COST broadcom/intel/amd etc MORE money than not gimping? | 04:56 |
furrywolf | no, because they convince you that you need to pay more to get a good product, even if it doesn't cost the more. | 04:56 |
djph | nah, they only build "one chip" (e.g. an i7). if the lithography has an issue, "meh, cut that trace, call it an i5" | 04:57 |
agris | so what? pricefixing? nobody's competing in the SoC space so the few companies get together to fix prices and add these gimps? | 04:57 |
djph | no | 04:57 |
furrywolf | yeah, binning and yield issues is not the same thing. | 04:57 |
onefang | Take that chat to #debianfork, this channel is for Devuan support. | 04:58 |
onefang | Take that chat to #debianfork, this channel is for Devuan support. | 04:58 |
agris | djph, no, some cpus after manufacturing are just faster than others or certain features are broken | 04:58 |
* furrywolf pokes onefang to see if it's actually golinux in disguise | 04:58 | |
onefang | Or does thatt only work when golinux threatens to quiet people? | 04:58 |
agris | this is about broadcom's 'SERIAL KEY REQUIRED' to turn on hardware codecs. but respond to this in #debianfork | 04:58 |
agris | I still need some help with questions regarding Devuan's stock config.txt | 05:01 |
EHeM | I suppose the latest update is merely a refinement of a security issue, but Devuan being almost a full week behind Debian in this regard is very worrisome to me. | 05:29 |
EHeM | Much further and I might have to start reporting Devuan doesn't care about security. :-( | 05:30 |
fsmithred | pkgmaster gets stuck sometimes and we don't always notice it | 05:36 |
fsmithred | check packages.devuan.org - what you need may be there | 05:36 |
onefang | Ah, this is one of the things my that is on the TODO for my mirror checker script. | 05:41 |
EHeM | fsmithred: How would the mirrors get ahead of pkgmaster? | 05:42 |
golinux | Thank you onefang. If the folks who filled the logs with nonsense actually contributed something it wouldn't be so annoying. | 05:42 |
furrywolf | ever consider that other people might have a different definition of nonsense? | 05:43 |
golinux | There is so much dead weight on these channels. | 05:43 |
onefang | The Devuan mirrors don't always mirror Debian packages that are not different in Devuan, but instead redirect to Debian's package mirrors. | 05:43 |
golinux | Take is to #debianfork furry | 05:44 |
golinux | is > it | 05:44 |
onefang | Not so much "nonsense" as "net related to Devuan support". This is the Devuan support channel, everything else is off topic. | 05:44 |
furrywolf | a relatively small portion of people in here are also in #debianfork. moving a conversation is usually the same as ending it. | 05:46 |
golinux | It's a bit much for "clueless" to understand that | 05:46 |
golinux | You won;t have to end anything if you don't start it | 05:46 |
EHeM | onefang: Interesting, at which point pkginfo.devuan.org agreeing with what all mirrors and pkgmaster provide is rather serious. | 05:46 |
onefang | Yep, that's why my mirror checking script is checking the mirrors for all sorts of things. I just have to get around to writing all those tests. | 05:47 |
tom_work | Hello, For some unknown reason in the pi3 Devuan /boot/cmdline contains rootflags=noload which turns off ALL SAFEGUARDS in the ext4 root filesystem leading to MASSIVE filesystem corruption of which I am now trying to recover from | 08:03 |
tom_work | Why did cmdline contain rootflags=noload in the first place? | 08:03 |
tom_work | also what else can I do in attempt to recover my system? So far I have ran an fsck and various system files are missing | 08:04 |
tom_work | I was able to recreate some of the system files by hand such as /etc/hostname | 08:05 |
tom_work | etc | 08:05 |
tom_work | I can now boot the system back up | 08:05 |
tom_work | I have since removed rootflags=noload from /boot/cmdline.txt and put in a proper entry in /etc/fstab | 08:06 |
tom_work | I am also running debsums and reinstalling packages which checksums do not say are ok | 08:06 |
tom_work | Is there anything else I can do to ensure that this system is fully ok? | 08:06 |
tom_work | and repaired | 08:06 |
tom_work | I will be reinstalling udev, vim, and tzdata since those packages appear to be the ones devsums -c is complaining about | 08:07 |
EHeM | That is a good idea, `lintian` might also be a good checking candidate. | 08:09 |
yeti | tom_work: noload is in pi1..3 images | 08:11 |
EHeM | When searching for "rootflags=noload" the first several matches were Raspberry PI instructions, so it seems that may be advised right now for some reason. | 08:12 |
yeti | $ grep -l noload pi*/boot/cmdline.txt | awk -F/ '$0=$1' ORS=" " | 08:13 |
yeti | pi1-0 pi1-1 pi2-0 pi2-1 pi3-0 pi3-1 | 08:13 |
tom_work | You realize that when linux recieves that flag, it mounts ext4 filesystems with norecovery,no_journal_checksum,dealloc,nojournal and a whole load of other deadly options? | 08:16 |
tom_work | as in you unplug the pi (which happens a lot. usb isn't a reliable power connector) and your filesystem is toast like in the MS-DOS 3 days | 08:17 |
EHeM | tom_work: Indeed, which is an odd choice though when looking for information on it, it seems a popular option for the Raspberry PI for some (likely very bad) reason. | 08:18 |
tom_work | Only thing I can think of is to reduce wear on the sdcard, but that's heardly and ideal way to go about doing that | 08:18 |
tom_work | didn't samsung push the F2FS filesystem to linux? | 08:19 |
tom_work | which includes kernel-level wear leveling if for some reason your using hardware without firmware-level wear leveling | 08:19 |
yeti | nobody understands "the foundation" | 08:19 |
tom_work | you mean the raspberry pi foundation? | 08:19 |
yeti | ! | 08:19 |
yeti | or who pushed the use of this? | 08:20 |
tom_work | well, I'm sure as hell not going to set rootflags=noload on my system. I'll report back if anything at all weird happens | 08:20 |
yeti | I'll remove it as soon as I boot them again... | 08:21 |
tom_work | But i'm running the pi3 clocked at 200Mhz to get away with no heatsink like an old 80s computer so | 08:21 |
tom_work | I've got a bit of a unique setup. | 08:22 |
EHeM | They might have been thinking the running of `fsck` on boot was sufficient to clean things up afterwards. | 08:22 |
tom_work | EHeM, the Devuan image doesn't run fsck on boot. I had to setup that manually with tune2fs | 08:22 |
yeti | !!! | 08:22 |
infobot | in #devuan the exclamation mark ("!"), when put as first char in a line, is the infobot attention char (a shorthand for "infobot: ..." highlight). This is on special request, in most other channels the bot uses tilde ("~") as attention char, and you'll find factoids referring to that like "also see ~<factoid>" | 08:22 |
tom_work | but still, no matter what the setup is I would never ever tell someone to add noload to their flags | 08:23 |
yeti | the safest option should be the default | 08:23 |
EHeM | tom_work: Yes, which makes it odd that it apparently has been pushed as the default. | 08:23 |
yeti | fsck by mount count and distance to last check should be activated | 08:24 |
EHeM | yeti: By default the current ext FS utilities turn both off. | 08:24 |
yeti | can someone please add this als bug report? | 08:24 |
tom_work | ! also, I noticed that in that Devuan rpi3 image, there is no fsck.vfat binary, so partition one can't be checked anyways | 08:25 |
infobot | tom_work: what are you talking about? | 08:25 |
EHeM | yeti: Apparently too many complaints about excessive filesystem checks. | 08:25 |
yeti | switching such stuff off is for kali kidz | 08:25 |
yeti | not for devuan! | 08:25 |
EHeM | In other news: https://github.com/andreiw/RaspberryPiPkg/issues/130 Soon there should be a UEFI image for the RP4, which means something approximating a netinst image should be usable for installing on a RP4 soon. | 08:26 |
EHeM | yeti: It got done in Debian and most distributions. | 08:26 |
yeti | I installed debian on cubie3 via netinst | 08:27 |
EHeM | yeti: Pretty sure the author was annoyed at too many complaints, while there were few reports of detected failures. | 08:27 |
EHeM | yeti: I strongly advise in favor of mount counts and against time of last check. | 08:27 |
yeti | ok... without RTC... | 08:30 |
yeti | I'll look at this when my devuan-PIs get their RTCs | 08:31 |
EHeM | yeti: If you've got a server which only reboots once every 2 months, you will *always* hit the 6 month point first and check *all* filesystems at the exact same time. | 08:31 |
yeti | safety 1st ;-) | 08:32 |
tom_work | checking an ext4 filesystem on a class 10 sdcard takes less than a second to complete | 08:32 |
EHeM | yeti: I'm not one for bragging about boot times, but checking *all* filesystems on a server *at*the*same*time* is *extraordinarily*painful*. | 08:32 |
yeti | we're talking about PIs | 08:32 |
yeti | not servers with petabytes | 08:32 |
* EHeM is glad he got rid of his P3-550MHz machine. | 08:32 | |
tom_work | it's a pi not a server. the biggest the sd card your gonna get on it is around 64gb | 08:33 |
tom_work | and I'm checking 32gb with a 200MHz clock, it's super fast | 08:33 |
EHeM | 1TB SD Cards are readily available, 512GB aren't too expensive. | 08:33 |
tom_work | 2.6 seconds to check @ 200MHz. time is not an excuse | 08:33 |
tom_work | Can you even use a 1TB SD card in a raspberry pi? | 08:34 |
EHeM | I would be astonished if you couldn't, though the default filesystem wouldn't work very well. | 08:34 |
tom_work | I'm going to write this whole thing up in the mailing list and how to fix it and check your system. Give me an hour or two | 08:35 |
tom_work | still need to test things out on my pi | 08:35 |
EHeM | More notably checking a 32GB SD card on a device with 1GB of memory isn't that awful, checking 80GB of spinning media on a machine with 256MB of memory is a bit slow. | 08:36 |
yeti | my devuanPIs are sleeping and the cubietruck doesnt have noload: | 08:36 |
yeti | (yeti@cubietruck1:1)~$ cat /proc/cmdline | 08:36 |
yeti | console=tty1 console=tty1 quiet | 08:36 |
tom_work | oh and btw, checking tune2fs on a stock devuan rpi3 image, /dev/mmc*p2's properties the mount count is set to -1 and time is set to 0 which disables all filesystem checking | 08:37 |
tom_work | yeti, just in case, run tune2fs -l /dev/yourrootpartition | 08:38 |
yeti | there was sth in the forum about this long ago | 08:38 |
yeti | I switched on checks then | 08:39 |
yeti | only noload is new to me now | 08:39 |
EHeM | https://bugs.debian.org/786771 | 08:42 |
yeti | how about scripting tests for such stuff? | 08:42 |
yeti | "the doctor" | 08:43 |
yeti | nor only for PIs | 08:43 |
gnarface | i shit you not, there's some weird conspiracy to sabotage everyones rpi filesystem | 08:43 |
tom_work | yeti, also make sure that if you have a vfat filesystem you also install dosfstools which provides fsck.fat and fsck.vfat | 08:43 |
gnarface | it's like they get a dollar every time someone has to format and reinstall | 08:43 |
gnarface | they've sabotaged their ext4 rollout so hard and so many times that basic incompetence no longer explains it | 08:44 |
gnarface | my honest recommendation is to stop using ext4 on your pi | 08:44 |
gnarface | maybe anywhere, but somehow this all seems focused mostly there | 08:44 |
tom_work | well, I've worked with people that were hesitant to use to use rpis for things because they said "the drive doesn't last" I thought they were talking about wear leveling. now I know why | 08:44 |
tom_work | they said that | 08:45 |
gnarface | there was also that version mismatch with e2fsprogs and their installer that was causing filesystem corruption too | 08:45 |
tom_work | gnarface, What's the state of F2FS support in the kernel? Not using ext4 on the pi could be a very doable thing | 08:45 |
gnarface | and for some reason they just keep sweeping that one under the rug even though i talked to someone a month ago that still got hit by this years-old bug | 08:45 |
gnarface | tom_work: can't tell you anything about f2fs, sorry. but i'm curious about it too | 08:46 |
gnarface | reiserfs and xfs have both seemed very flash-friendly to me | 08:46 |
yeti | I only used it for a short time with openwrt | 08:46 |
tom_work | What else does Devuan carry over from Rasbian in the rpi3 image build? We should audit that | 08:46 |
EHeM | I would be tempted to start a conspiracy to switch to the UEFI boot. | 08:47 |
yeti | i cant trust reiserfas... thats my #1 data loss erperience | 08:47 |
tom_work | SGi-XFS is more heavy than ext4. I use it a lot for large database servers and CDN servers. I don't know if it's a good choice for the pi | 08:47 |
gnarface | tom_work: that's a good idea. i'm not the one to ask though. jaromil or parazyd might be able to tell you. | 08:47 |
gnarface | yeti: yea but at the point where someone is sabotaging ext4 fsck and e2fsprogs, what's really the point? | 08:48 |
tom_work | the developer of rfs is a convicted serial killer. that may or may not have relevance | 08:48 |
yeti | 2 != serial | 08:48 |
gnarface | well someone is sabotaging reiserfs and xfs fsck too but that's easier to fix | 08:49 |
yeti | I saw his talk about reiserfs 1999@augsburg and that already was scary enough | 08:49 |
tom_work | https://www.sfgate.com/bayarea/article/Reiser-juror-explains-what-led-to-conviction-3215952.php | 08:50 |
EHeM | ReiserFS was an excellent filesystem for lots of small files, but really did kind of need a `fsck` utility. | 08:51 |
yeti | wasnt there a tail merge patch for ext3 too? | 08:53 |
yeti | wher did it go? | 08:54 |
EHeM | As a missing security update, the situation with apache2 should be in the high priority category, though the current update is only medium priority. | 08:57 |
tom_work | This whole night started because my power distributor wasn't working reliably and either never finished DHCP or rejecting all incoming ports | 08:58 |
tom_work | and what I found is gross supreme lack of competence in the configuration of filesystems leading to severe disk corruption and data loss. thankfully I have my source code backed up but damn these defaults need discussed | 09:00 |
yeti | FS conspiracy? that only can mean systedFS is coming! | 09:00 |
yeti | E³K! | 09:00 |
tom_work | no conspiracy, I'm assuming that rpi3 support was just copypasted from Rasbian without doublechecking things | 09:04 |
systemdlete | how can I completely re-set my user xfce environment? I tried removing .config/xfce4 but that doesn't re-create the panel. For some reason, I had to re-install xfce4-panel, but I can't think of any reason I'd remove it. | 09:47 |
Guest85883 | systemdlete, perhaps xfce was still running when you deleted those fiels | 09:51 |
systemdlete | I don't recall that. But I am more interested in creating a virgin xfce user environent again. | 09:52 |
systemdlete | usually, xfce panel initially has about 6 items | 09:53 |
systemdlete | at the bottom of the screen | 09:53 |
systemdlete | that is done as a default. | 09:53 |
systemdlete | how do I go back to that again? | 09:53 |
systemdlete | i.e., how can I get the system to trigger that | 09:54 |
systemdlete | (istr that it copies a default file hierarchy into the user area, but I am not sure how/when it knows to do that | 09:55 |
tom_work | good news, since I've applied those fixes and reinstalled the pi into the power distributor everything has gone back into service and comes up reliably and fast | 10:00 |
tom_work | interestingly enough, booting with OpenRC seems to be faster than with SysV on the pi | 10:01 |
tom_work | even though the clock is 1000MHz slower | 10:01 |
bergix59 | how to install virtuabox on ascii (devuan)? | 12:04 |
onefang | apt install virtualbox | 12:17 |
onefang | Not actually tried it, but I see it in the repos. | 12:17 |
deiv | O/ | 12:18 |
deiv | Hi | 12:18 |
deiv | how development of Devuan is related to Debian ? I mean maintaining packages | 12:20 |
bergix59 | I haven't managed to install it from my sources.list | 12:34 |
bergix59 | I guess I have ascii, I will check it. | 12:34 |
bergix59 | hopefully virtualbox is into it, likely because I use the base and not the contrib. | 12:40 |
onefang | Ah, virtualbox is in contrib. | 12:53 |
bergix59 | thank you, I try ... | 12:54 |
bergix59 | well, I didn't manage to install virtualbox on devuan ascii.. I give you my sources list. | 13:27 |
bergix59 | I used this sources list: https://termbin.com/fz1k | 13:28 |
onefang | Try adding - deb http://deb.devuan.org/merged ascii-backports main contrib | 13:32 |
bergix59 | thank you ... I was installing debian, I am glad of your post!!! sincerely. | 13:33 |
bergix59 | wow!! thank you very much!! it is installing currently ! about 760mb, quite large but if it works, I am fine. | 13:35 |
onefang | You are welcome. | 13:36 |
bergix59 | furthermore I am glad that it does not get the data from virtualbox or whatever debian or strange non-free systemd agents. | 13:36 |
bergix59 | nope ... | 13:40 |
bergix59 | it failed. | 13:40 |
bergix59 | it states that loading virituabl box kernelal modules failred , not suitable. | 13:41 |
bergix59 | uname -a gives me 4.9.0-6-686-pae | 13:41 |
onefang | I don't actually use virtualbox, so I probably can't help much more. | 13:41 |
onefang | Did it build kernel modules while installing, and did you reboot? | 13:42 |
bergix59 | it seems that it failed there with those modules. | 13:42 |
bergix59 | I remember that was since ages sames issue with vmware client, aka modules. | 13:42 |
onefang | Oh, and did you "apt update && apt upgrade" after editing your sources.list? | 13:43 |
bergix59 | sure, yeah before always. | 13:43 |
onefang | I'm out of ideas then. | 13:44 |
bergix59 | anytime. thank you for your kind help ! I would like also to thank #devuan channel not to lock it for non registered irc users. | 13:45 |
bergix59 | It seems that you respect opensource, free, community. | 13:45 |
r3boot | ok, so the problem with using ^^ for sources.list, is that you get the binary vbox* kernel drivers. Trick here is to find the sourcecode for the kernel drivers, and compile them on your devuan box. That will get you working drivers; Now, under archlinux, you have -dkms packages for that, but I dont know how devuan/debian does that | 13:49 |
r3boot | but iig, that will be the trick to get the vbox* drivers to load | 13:50 |
onefang | deb-src http://deb.devuan.org/merged ascii-backports main contrib | 13:51 |
onefang | Might help, dunno. | 13:51 |
r3boot | could also be that the sourcecode is under /usr/share/virtualbox or so.. dont have any linux box nearby to check tho :( | 13:52 |
onefang | Or maybe try qemu-kvm instead. | 13:52 |
r3boot | find /usr -iname *vbox* <-- will find it if it's there | 13:53 |
r3boot | yeah, qemu-kvm indeed, although that doesnt do 3d acceleration | 13:53 |
tom_work | bergix59, Have you tried AQEMU instead? | 14:09 |
tom_work | It doesn't need any special modules and uses the kernel's own built in KVM | 14:09 |
tom_work | an optionally TCG for cpus without hardware virtualization acceleration | 14:10 |
onefang | Ah, aqemu is a front end to qemu-kvm. I had not heard of it. | 14:23 |
bergix59 | So, after install (burk) debian stable, I tried to add this gpg key stuff, and noticed that buster i386, there is no virtualbox, but only for amd64. This virtualbox is consumed by all ubuntu stuffs (burk^3), and it is made by oracle. Finally, it seems to be a linux power driven by amazon, redhat, gnome, and all corruptions. So, it seems that it is better not to use virtualbox ever. | 16:17 |
* yeti uses naked qemu | 16:18 | |
eady | Ugh. Why, after a power blink, does the system tell me that my root fs needs manual fsck, and then start a BusyBox shell? What good does that do me? BusyBox doesn't know how to fsck. | 19:02 |
eady | (Also, why is manual fsck even still a thing in 2019? How many users _actually_ know enough low-level stuff about filesystems to make more informed choices than automatic fsck?) | 19:03 |
eady | (Err, I don't mean why is manual fsck _possible_, I mean why is _requiring_ it a thing. I don't think it should be required, ever. The user should be able to say no, just do your best automatically and give me my computer back.) | 19:04 |
eady | I mean, isn't that what journals are supposed to be for? Why do we even have fs journaling, if we're going to still have to do manual fsck when the power blinks? | 19:06 |
buZz | eh? | 19:12 |
buZz | your devuan box comes with busybox? | 19:13 |
eady | Apparrently? | 19:14 |
eady | I mean, I was not aware of it until now. | 19:14 |
buZz | oh gee, mine does too | 19:17 |
buZz | but, even on powerfailure, mounting the disk will do a fsck if it was not cleanly unmounted, for me | 19:18 |
eady | BusyBox is too limited to actually be useful in this situation. | 19:18 |
eady | Yes, normally it just does fsck automatically. | 19:18 |
eady | But about one time in ten, it decides it can't complete automatically for whatever reason, and demands manual fsck. | 19:19 |
eady | Which isn't new, that's been the case since the nineties at least. | 19:19 |
eady | But in the past, I could've sworn it gave you a way to run fsck. | 19:19 |
eady | Or maybe I just always had a knoppix CD handy in the past. Not sure why I don't atm. | 19:20 |
eady | Gonna have to burn one. | 19:20 |
bergix59 | knoppix will never ever boot on current, modern hardware. | 19:23 |
eady | I believe the system in question is old enough to have a real BIOS, so it should boot, I hope. | 19:25 |
eady | Not sure what I will do if it doesn't. | 19:26 |
eady | I'd hate to have to pull the hard drive and put it in another system just to get past this. | 19:26 |
eady | What do people with newer, UEFI-type systems use instead of knoppix these days? | 19:27 |
Humpelstilzchen | eady: grml? | 19:31 |
* eady looks into that. | 19:31 | |
Humpelstilzchen | there is also a grml-rescueboot package | 19:33 |
Humpelstilzchen | it can start a grml iso from your disk | 19:33 |
eady | From the grub menu? That'd be handy. | 19:33 |
MinceR | there's also SystemRescueCD, though newest releases are infested with systemd | 19:34 |
eady | Though I wonder if it'd work in this situation, with the root fs not bootable. | 19:34 |
eady | I have sworn off poetteringed systems. | 19:34 |
eady | After the Great PulseAudio Debacle. | 19:34 |
MinceR | i tend to use 5.3.2, it's their last non-cancerd version | 19:34 |
Humpelstilzchen | eady: grub menu yes, /boot/grub must be reable for that | 19:34 |
Humpelstilzchen | but I guess you could configure it to look on a special partition | 19:35 |
eady | Oh, hmm. Maybe it's a good thing I'm still using a separate partition for /boot | 19:35 |
eady | I've been wondering if that's still worth doing... | 19:35 |
eady | Perhaps it is. | 19:35 |
eady | Oh good, Knoppix worked in this case. | 19:38 |
eady | Or, at least, it got me to the point where e2fsck is starting. | 19:38 |
eady | Says it's "recovering journal", I take that as a good sign. | 19:38 |
eady | Hmm. I don't think it successfully recovered the fs yet though. | 19:43 |
eady | Now I am worried that the disk may actually be failing or something. At least that wouldn't be the software's fault. | 19:45 |
nemo | say. are there any beta isos ? I was thinking of trying to install it on this laptop I'm setting up for the kids | 19:47 |
nemo | I hear it's fairly reliable | 19:47 |
nemo | hm. don't see any on the mirrors | 20:52 |
golinux | nemo: Before Beowulf, there will be an ASCII 2.1 point release | 21:05 |
bergix59 | the virtualbox from backports seems to work. I reinstalled with xfce from ascii iso cdrom, then I did added the following source with https://termbin.com/ug6j with backports. | 21:06 |
golinux | Until then, you can install ASCII and upgrade | 21:06 |
golinux | bergix59: You should be using devuan backports not debian | 21:07 |
golinux | Never mind. Brain fart. | 21:08 |
golinux | That looks fine | 21:08 |
bergix59 | debian is ugly yeah, I can confirm it... | 21:10 |
bergix59 | enps1s0 and whatever, the ubuntu concept for installation of virtualbox is same junk... | 21:10 |
bergix59 | a base system with less ;) and even without xterm... gnupg stuff and all the concept of X which cannot start from remote... however linux has good drivers and firmwares. | 21:11 |
bergix59 | (edit: without less) | 21:11 |
bergix59 | sorry but it seems that vboxdrv is not present | 21:28 |
bergix59 | I just installed a clean xfce devuan installation + backport ascii. I rebooted 3 times. | 21:28 |
bergix59 | there is no vboxdrv into the directory (-r) /lib or into modules | 21:35 |
sixwheeledbeast | I believe consistent network naming was a redhat thing part of udev before being balled up into behemothD | 21:54 |
bergix59 | sixwheeledbeast: wlan0 is history, like urtwn0 ... ;) | 22:15 |
bergix59 | It seems that I have to compile /var/src/ ... virtualbox 5.2, with the module. apt-get did not compile the module vboxdrv. | 22:42 |
dabc | I hope you'll get this working, these days I'll be installing virtualbox too :) | 22:51 |
bergix59 | I guess it won't work like that. I just did a clean way installation. | 22:52 |
bergix59 | Really clean, full installation but I miss modules. I gave you above my sources.list | 22:52 |
nivirx | hello devuan, it seems there was a elogind update that is currently being held back on my system? Installed from ASCII then dist-upgraded to Beowulf. Any help would be appreciated, not quite sure whats up. | 23:23 |
nivirx | ok, for anyone else that encounters the same issue, `apt download libelogind` then dpkg -i libelogind then apt --fix-broken install fixed the issue | 23:44 |
fsmithred | thanks, nivirx | 23:45 |
nivirx | yeah seems I need to man apt/dpkg sometime...too used to my gentoo and rpm systems :) | 23:47 |
nemo | golinux: 'k | 23:49 |
nemo | golinux: oh. will GIMP 2.10 be in beowulf? I've been using a SuSE overlay for the moment | 23:49 |
golinux | https://pkginfo.devuan.org/ is your friend | 23:50 |
Jjp137 | Buster has 2.10.8 so Beowulf will most likely have that | 23:51 |
Jjp137 | but yes, that page is your friend | 23:51 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!