awill | hello, i'm having some issues with apt on my devuan install. is this the right place to ask? | 03:29 |
---|---|---|
fluffywolf | yep! | 03:30 |
awill | great, what could help with diagnosing my issue? | 03:30 |
fluffywolf | well, first, saying what the issue is... | 03:30 |
awill | right, derp :p | 03:31 |
awill | any time i try to perform any actions that request deb.devuan.org, there's a 75% chance it will error out and fail to connect to the server. be it update or installing a package, it fails most of the time | 03:31 |
awill | both apt and apt-get experience the same issue | 03:31 |
onefang | Is there a specific IP for deb.devuan.org when you get that error out? | 03:32 |
fluffywolf | onefang can probably help with that a lot better than I can. | 03:33 |
onefang | deb.devuan.org is a DNS round robin, you'll get a different IP pointing to a different server each time. | 03:33 |
awill | of course when i try to replicate the problem it works fine, ugh | 03:33 |
fluffywolf | my first thought would be you have non-apt-related network isues... | 03:33 |
onefang | Or that. | 03:33 |
awill | potentially? i have two network cards in my pc and both have worked fine in other distros, this is the only instance where i'm having network issues | 03:34 |
onefang | We use https://sledjhamr.org/apt-panopticon/results/Report-web.html to monitor the package servers, you can check the IP you get against that to see if that particular server is having issues at that time. | 03:34 |
onefang | You can also use https://pkgmaster.devuan.org/mirror_list.txt to pick a particular server instead of deb.devuan.org if you prefer. | 03:35 |
onefang | But failing 75% of the time means the problem is likely closer to your end of the network. | 03:36 |
onefang | Coz if the servers where failing that often, EVERYONE would be bitching at me to fix it. lol | 03:36 |
awill | true, i suppose i could try hotspotting to rule out my isp | 03:37 |
awill | things seem to be working after rebooting router, thanks for the help onefang and fluffywolf! | 03:43 |
fluffywolf | well that's an easy fix. :) | 03:45 |
fluffywolf | are you using bittorrent or other peer-to-peer applications? | 03:45 |
awill | not that i know of, no | 03:47 |
fluffywolf | k. some cheap routers often fail badly when you use such things, from all the connections. | 03:47 |
awill | it's my isp provided router, can't use anything else sadly | 03:49 |
awill | no idea what the quality is like | 03:49 |
fluffywolf | terrible, I would assume. lol | 03:49 |
onefang | Gone #devuan-offtopic now. | 03:51 |
gry | < awill> it's my isp provided router, can't use anything else sadly | 04:41 |
gry | can you buy your own? | 04:41 |
gnarface | in the US it's not illegal for the ISP to require you use the hardware they tell you to use | 04:44 |
gnarface | it's not even illegal for them to charge you for it on the way in and the way out | 04:45 |
gnarface | the best you can do is put another firewall behind it | 04:45 |
gnarface | (that's still not illegal, yet, though it is also not illegal for them to tell you it's illegal so lots of people fall for this scam) | 04:45 |
gnarface | he's already gone though, gry | 04:48 |
onefang | No idea if they was even in USA. | 05:14 |
gnarface | well, it was just as an example, but it's a safe bet because it usually doesn't come up otherwise | 05:18 |
fluffywolf | the cable isp here recently changed ownership, now says you have to use their equipment... I should hopefully be getting them next year... will probaly put their modem/router in a wifi-proof box, then plug my own router into it. | 05:29 |
systemdlete | I have 2 brand new 2TB SATA hard disks (mechanical, not ssd) which are connected to a brand new mainboard. Booting from the daedalus live iso, fdisk -l shows that there are no partitions on either drive. | 09:41 |
systemdlete | But /dev/sda[1-4] and /dev/sdb[1-4] exist. Forgive me if this is some elementary thing I have forgotten about. | 09:42 |
systemdlete | In most cases I've ever encountered, those device files do not get created until there are partitions on those disks recorded with the kernel. | 09:43 |
systemdlete | Here, the kernel seems to think these partitions already exist, yet fdisk -l shows no partitions on either drive, and even blkid does not know of them either. | 09:44 |
systemdlete | So take me back to school if this is something I should already know. | 09:44 |
systemdlete | MB is Asrock 970M with an FX8350 and 32GB DDR3. | 09:45 |
n4dir | where does daedalus live iso show the 4 partitions? You seem to say fdisk -l doesn't show them. | 09:46 |
systemdlete | <systemdlete> But /dev/sda[1-4] and /dev/sdb[1-4] exist. | 09:46 |
systemdlete | i.e., ls /dev/sd[ab]* | 09:47 |
n4dir | i see, they do exist, but on the live iso, at least fdisk -l doesn't know about them? | 09:48 |
n4dir | i'd have a quick look at gparted too, without expecting much insight from it, but it won't hurt | 09:48 |
systemdlete | n4dir, are you saying it is just getting the info from the iso image? | 09:48 |
n4dir | systemdlete: no, i am saying it is confusing, i wouldn't expect that either | 09:49 |
n4dir | but first i didn't fully understand the facts, that is why i asked | 09:49 |
systemdlete | gparted is not in the path, but parted is. Parted print shows no partitions | 09:49 |
n4dir | yeah, no need to install gparted, as said it probably won't help much anyway | 09:50 |
n4dir | external or internal disks? If exernal you could unplug them, plug them back, then look at dmesg | 09:50 |
systemdlete | n4dir, but you brought up an interesting point. So that got me to thinking that maybe the /dev/sd* files are from the image | 09:50 |
n4dir | i for one don't know too sure what happens to /dev/* when. I usually just use fdisk | 09:51 |
systemdlete | they are internal. Sadly, my multi-bay unit came in missing a piece, so I am waiting for it (Newegg switched product on me). So to remove the drive and re-attach it means opening the box (front AND back panels, etc) | 09:52 |
n4dir | yeah, that is a bit too much | 09:52 |
systemdlete | I am so pissed, too! | 09:52 |
n4dir | i can relate | 09:52 |
rrq | if "brand new" they might be single partition with ntfs or not partitioned at all | 09:52 |
systemdlete | That they swapped the product on me is one thing. That it came incomplete is another. | 09:53 |
n4dir | if you have other live isos around, i'd have a look what they say | 09:53 |
n4dir | got a link for the live iso you used? I could quickly start it in qemu | 09:53 |
systemdlete | I do have daedalus install iso... | 09:53 |
n4dir | i wouldn't expect much insight from that either | 09:53 |
systemdlete | devuan_daedalus_5.0.0_amd64_minimal-live.iso | 09:54 |
n4dir | oh, i meant the link. I am not too good with devuan website and web pathes | 09:54 |
systemdlete | well, no, actually you have a good idea. Maybe other iso images are configured differently. | 09:54 |
systemdlete | oh | 09:54 |
systemdlete | hold on | 09:54 |
n4dir | would take a bit of time, first download it, then boot, but it may (or may not) help | 09:55 |
systemdlete | https://mirror.leaseweb.com/devuan/devuan_daedalus/minimal-live/devuan_daedalus_5.0.0_amd64_minimal-live.iso | 09:55 |
systemdlete | (I appreciate your help!) | 09:55 |
n4dir | ok, give me a few | 09:55 |
n4dir | not much help, sorry, but as long no one with a clue has a say .... | 09:56 |
systemdlete | but you can download from any of the mirrors. I just picked the first one | 09:56 |
systemdlete | I'll try booting other images, just for comparison. | 09:56 |
rrq | you can run installer-iso and inspect drives instead of installing | 09:58 |
fsmithred | systemdlete, did you just create the partitions that fdisk can | 10:00 |
n4dir | systemdlete: login username and password are? | 10:00 |
fsmithred | can't see? | 10:00 |
fsmithred | if so, run partprobe | 10:00 |
fsmithred | devuan:devuan root;toor | 10:01 |
n4dir | thanks | 10:01 |
systemdlete | geez, I didn't have to enter a user or password. | 10:02 |
systemdlete | fsmithred, I haven't done ANY partitioning on either disk. | 10:02 |
n4dir | systemdlete: at least here, in qemu, it does see the partitions with fdisk | 10:02 |
systemdlete | I boot up, obtain root permissions if I don't already have them, and run fdisk -l | 10:02 |
systemdlete | n4dir, huh. | 10:02 |
systemdlete | but are those blank drives? | 10:03 |
systemdlete | Well... | 10:03 |
systemdlete | I guess what I mean is, you know that the disk images you passed to qemu are essentially virgin? | 10:03 |
n4dir | no, they have a partition, and they get seen. Looks like i still didn't understand the facts of your situation | 10:03 |
systemdlete | (no partitions etc) | 10:04 |
systemdlete | I did say these were brand new disks. | 10:04 |
systemdlete | Also, I wonder if running under qemu vs physical hardware might give different outcomes. | 10:04 |
systemdlete | n4dir, can you re-try your experiment with completely empty new disk images? | 10:05 |
systemdlete | (they don't have to be huge, just enough to test) | 10:05 |
n4dir | i just created it | 10:05 |
systemdlete | did you partition it? | 10:05 |
rrq | note that one can create device nodes like /dev/sda1 etc without actuallly having partitions | 10:06 |
n4dir | yup, the just created new "disk" without being partitioned only gets shown as "/dev/sdb" with fdisk -l | 10:06 |
systemdlete | rrq, right. | 10:06 |
n4dir | the already exisiting and partitioned one gets shown as /dev/sda and /dev/sda1 | 10:06 |
systemdlete | n4dir, which is what I see also | 10:07 |
systemdlete | which we would expect also for a previously partitioned disk | 10:07 |
systemdlete | thanks, n4dir | 10:07 |
n4dir | same for /dev/* | 10:07 |
systemdlete | at least the results are consistent, even if confusing. | 10:07 |
systemdlete | rrq, why would those nodes be created if there are no partitions? | 10:07 |
n4dir | well, the hardest part was switching between german and english keyboard layout. not hard, confusing. | 10:07 |
systemdlete | oh, sorry about that. | 10:07 |
systemdlete | I sometimes forget about the language barriers | 10:08 |
rrq | it's possible (in tehory) that the live iso comes with prepolated /dev/sdaN | 10:08 |
systemdlete | rrq, that was one of the scenarios I considered also. But again, that really shouldn't be the case, should it? | 10:08 |
n4dir | been ages i ran in that problem, if a GUI is involved, i just setxkbmap de. On a TTY it is a bit strange :-) Anyway, back to topic | 10:08 |
n4dir | rrq: yeah, but the same iso for me it doesn't have it | 10:09 |
systemdlete | n4dir, np. I appreciate your assistance | 10:09 |
rrq | try "mount -t devtmpfs xxx /dev" to see what the kernel thinks | 10:09 |
systemdlete | wait n4dir didn't you just tell me you are seeing those nodes? | 10:09 |
systemdlete | (now I'm wondering if we are saying/seeing the same things) | 10:09 |
n4dir | I see it the way it really is: /dev/sda ; /dev/sda1 and /dev/sdb (no /dev/sdb1 ) | 10:10 |
systemdlete | rrq, great idea, thanks. HOld on | 10:10 |
systemdlete | n4dir, and sdb is the "new" disk, right? | 10:10 |
n4dir | corrcect, sir | 10:10 |
systemdlete | ok | 10:10 |
systemdlete | so what did you mean by your remark to rrq then? That confused me. | 10:11 |
n4dir | he said at least in theory those "nodes" might already be in /dev on certain live isos | 10:11 |
systemdlete | rrq, wrong fstype, bad superblock, etc message (can't copy-paste it, something like that though) | 10:12 |
systemdlete | rrq, I am nearly 100% certain that the partitions really don't exist. It is just the /dev file system that seems to think they do | 10:12 |
n4dir | on a side note i got those USB sticks which i always see as /dev/sdb, no partitions no nothing, but if i boot them "mount /dev/sdb /media/stick", they work just fine. Also weird | 10:12 |
systemdlete | and like you suggested, probably these are "burnt" into the iso image | 10:12 |
rrq | mmm are proc and sysfs mounted... yes I would trust fdisk more then device nodes | 10:13 |
systemdlete | n4dir, not too weird, actually. I have that scenario sometimes also. If the entire device is formatted with a file system, you can still mount it. file systems don't HAVE to be in partitions | 10:13 |
n4dir | i recreated it several times. The stick is 10 years or older, so i don't really think much about it. Only use it to move config files from one machine to the other | 10:14 |
fsmithred | I don't think those nodes are populated by the live iso. Some are, but not /dev/sd* | 10:16 |
systemdlete | fsmithred, then where are they coming from? | 10:16 |
systemdlete | any idea? | 10:16 |
rrq | dev creates them from /sys | 10:16 |
fsmithred | I don't know. I didn't see them in a refracta no-X iso. I don't have a devuan minimal-live available to test right now. | 10:16 |
fsmithred | I need to go back to sleep. It's 4am here. | 10:17 |
systemdlete | fsmithred, btw, I just tried partprobe (booted under chimaera desktop live) and it made no difference. The nodes are still there, and fdisk -l is still reporting no partitions | 10:18 |
rrq | *udev* | 10:18 |
systemdlete | fsmithred, night. thanks | 10:18 |
systemdlete | rrq, so /sys might be hardcoded into the iso? | 10:18 |
rrq | well, /sys will hold sysfs from the kernel, and /sys/block/* are the block devices it sees | 10:19 |
systemdlete | (sorry if my questions are dumb) | 10:19 |
systemdlete | then it is sounding like /sys is not being built dynamically, but is hardcoded somehow and the kernel is pulling that info? | 10:20 |
rrq | then you would have a /sys/block/sda/sda1 "derectiory" if the kernel sees such a partition | 10:20 |
rrq | "directory" | 10:20 |
systemdlete | ok, let me look... | 10:20 |
rrq | you need to have it mounted... mount -t sysfs sysfs /sys | 10:21 |
systemdlete | yep. They are there under /sys/block/sda, all 4 partitions | 10:21 |
rrq | and to mount, you need proc mounted | 10:21 |
rrq | interesting | 10:21 |
systemdlete | I'm pretty sure they are, but I'll double check. | 10:21 |
systemdlete | indeed, they are mounted... | 10:23 |
rrq | and right now you have 3 drives ? (including the live iso) | 10:23 |
systemdlete | atm, I have booted from a usb stick, and it gets assigned drive sdc | 10:24 |
systemdlete | the other 2 are the 2TB sata drives, unpartitioned (to the best of my knowledge) | 10:24 |
systemdlete | blkid and fdisk report null results (no partitions) for those drives | 10:24 |
rrq | right; .. fdisk -l says so as well.. hmm | 10:25 |
systemdlete | do you agree this is a bit weird? | 10:25 |
systemdlete | or at least, inconsistent? | 10:25 |
rrq | yes. | 10:25 |
systemdlete | And I've installed other distros (included earlier devuans going back to ascii) and never noticed this before, but perhaps I've never installed to virgin disk drives before, idr. | 10:26 |
rrq | fdisk recognises a fair few partition formats... is this one it doesn't recognize? ... which kernel version do you have | 10:26 |
systemdlete | rrq, whatever comes on the iso. | 10:26 |
systemdlete | but there shouldn't be any format at this point. | 10:27 |
rrq | is there a "gdisk -l" on the iso? | 10:27 |
systemdlete | these are new drives. The only thing I've done with them is run badblocks to make sure Newegg didn't send me duds | 10:27 |
systemdlete | hold on | 10:27 |
systemdlete | gdisk says it did not find any partition tables on the disk (I only tried one) | 10:28 |
rrq | ok. I'd suggest you run fdisk to define a partition, then reboot to see if the oddity vanishes | 10:29 |
systemdlete | I am 100% sure that will work. | 10:29 |
systemdlete | but I can do that. | 10:30 |
rrq | and then put this story into the "never happened" pile :) | 10:30 |
systemdlete | I will do it to ONE drive | 10:30 |
systemdlete | rrq, why would I do that? | 10:30 |
systemdlete | It DID happen. | 10:30 |
systemdlete | It is Happening. | 10:30 |
systemdlete | rrq: I think maybe the ISO spinners need to know about this so it can be fixed. It's minor, I know, but incorrect. We should not be shipping ISOs with these sorts of issues. | 10:31 |
rrq | right. before "fixing it" could you run an installer-iso to see if it gets confused as well? | 10:32 |
rrq | run "Install" (not expert) and then use Alt-F2 at the first dialog | 10:33 |
rrq | then check /dev | 10:33 |
systemdlete | I haven't actually created a partition yet, but get this. I run fdisk /dev/sda and it started to run. | 10:33 |
rrq | ? | 10:34 |
systemdlete | It printed the first paragraph about changes remaining in memory etc. But it did not get to the 2nd paragraph, instead it errored out | 10:34 |
systemdlete | said it couldn't open /dev/sda !!! | 10:34 |
systemdlete | BUT | 10:34 |
systemdlete | I ran it a second time, and it is prompting me for a command. | 10:34 |
systemdlete | wtf? | 10:34 |
systemdlete | OK, let me try installer iso as you suggest... | 10:35 |
systemdlete | (have to switch back and forth between machines tied to the same kb and video) | 10:35 |
systemdlete | is netinstall ok for this? | 10:37 |
n4dir | sure | 10:37 |
n4dir | as far i can tell, and i am rather sure, the installer really is the same, not matter which installation iso | 10:37 |
rrq | (yes) | 10:38 |
systemdlete | rrq: same thing | 10:39 |
systemdlete | still seeing /dev/sd[ab][1234] but fdisk -l shows no partitions | 10:40 |
rrq | ok; I'll try to chase it... sometimes a "destructive" badblocks can do things to MBR which might confuse the kernel... we'll see if I can recreate it | 10:41 |
systemdlete | hmm. do brand new disks come with an MBR? | 10:42 |
rrq | well the bytes are there | 10:42 |
systemdlete | ok... | 10:42 |
systemdlete | and that would explain the 4 partitions! DOS mbr. right... | 10:43 |
systemdlete | and badblocks scrambled that | 10:43 |
systemdlete | I think we have found the enemy, and it is us. I mean, me. | 10:43 |
systemdlete | but still... | 10:44 |
systemdlete | udev finds 4 partitions in an MBR, even though fdisk can't seem to do the same. Of course fdisk and friends are utilities, not kernel, so | 10:45 |
rrq | does "blkid" enumerate the pertitions? | 10:45 |
systemdlete | no, it shows nothing about them | 10:45 |
systemdlete | the grand irony of all of this is, I think, the fact that in my first professional job, my first task was to work on the dos disk partitioner for an operating system. | 10:46 |
systemdlete | So I should have remembered that... but that was in... uh... 1985? | 10:47 |
systemdlete | The engrams in my brain have clearly deteriorated over time. | 10:47 |
n4dir | tell me about it | 10:48 |
rrq | hmm "hidden partitions" | 10:48 |
systemdlete | so udev detects them, but all these utilities do not ( all the ?disk utils and blkid) | 10:53 |
systemdlete | well, I think I've taken up enough space and time here tonight. Time for bed. I'll work on this tomorrow. | 10:53 |
systemdlete | thanks to n4dir and rrq and fsmithred for your help. Always appreciate it. | 10:54 |
systemdlete | (I have a dr appt in the morning, need to get sleep) | 10:54 |
rrq | night | 10:55 |
systemdlete | gn | 10:55 |
n4dir | :-) | 10:56 |
al1r4d | Ge! | 12:25 |
ted-ious | systemdlete: Did you figure out why your disks had partitions on them? | 23:23 |
ted-ious | I wonder what the smart data says for how long the disks have been powered on? | 23:24 |
ted-ious | Maybe they were used by something else before you got them. | 23:24 |
fsmithred | ted-ious, he said that badblocks messed with the MBR | 23:30 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!