SuaveDandy | Are the spaces before ; something mandatory tho? | 00:01 |
---|---|---|
SuaveDandy | Not familiar with bash scripting | 00:01 |
fsmithred | no, not mandatory there | 00:01 |
SuaveDandy | I'm used to be a programming student so I thought… | 00:02 |
fsmithred | the spaces inside [ ] are mandatory | 00:03 |
SuaveDandy | Tested it. | 00:03 |
SuaveDandy | Echoing it does not. | 00:03 |
fsmithred | no output means it's not a block device | 00:04 |
fsmithred | you would need to comment out some lines in the script to skip that test and the next one | 00:04 |
SuaveDandy | What script are you referring to? | 00:05 |
SuaveDandy | The one that mason has? | 00:05 |
fsmithred | no. /usr/bin/refractainstaller | 00:06 |
SuaveDandy | Oh. | 00:06 |
SuaveDandy | What lines tho? | 00:07 |
fsmithred | hang on | 00:07 |
SuaveDandy | I'm having a notepad ready. | 00:08 |
fsmithred | I'm going to paste it for you. Will have a link in a minute. | 00:14 |
SuaveDandy | You're pretty fast. Thought you'll be searching for an hour or two seeing how big of a work you made on that script. | 00:16 |
fsmithred | https://paste.debian.net/1163235/ | 00:17 |
fsmithred | been working on that script for 10 years. I kinda know my way around it. | 00:18 |
fsmithred | if you can stand to look at the extra output, run it as 'refractainstaller -d' so you get a more verbose log. | 00:19 |
SuaveDandy | fsmithred: Thanks. Will add it to a note and do all that later today. I need to get some sleep. | 00:28 |
SuaveDandy | fsmithred: I think if you added the option to install directly to /mnt to the installer, it would help others who'll run into the same problem. | 00:32 |
SuaveDandy | Because oh, boy, way it a bumpy ride. | 00:32 |
SuaveDandy | *was | 00:32 |
SuaveDandy | That was my choise to install Devuan on ZFS, I guess. | 00:33 |
rennj | with zfs on root now you all need is the bootadm/beadm. and updates and reverting become nothing but zpool snapshots | 01:37 |
rennj | https://pasteboard.co/Hhq8gOv.png 1000pkg updated..but i could always revert | 01:39 |
rennj | diff between releases | 01:39 |
rennj | freebsd i think has beadm ported | 01:41 |
rennj | wonder if ubuntu has something like that..since they got the zfs kernel module | 01:42 |
rennj | https://www.freebsd.org/cgi/man.cgi?beadm | 01:43 |
mason | rennj: They have bectl in base and beadm in ports. | 01:44 |
systemdlete2 | libsnmp30 depends, in part, on libsensors5? Is this so mail servers can detect if a message is hot and get special priority? | 04:47 |
systemdlete2 | Tell the truth. | 04:48 |
furrywolf | is that commentary on excessive dependencies? :P | 04:49 |
systemdlete2 | I have no idea, actually. | 04:49 |
systemdlete2 | Anyone? | 04:55 |
systemdlete2 | I mean, I've got plenty of diskspace, so there's no reason for me NOT to download every last package in the repo. Then I would never need to worry about missing dependencies, although upgrades would take a lot longer. With Covid-19 keeping me grounded, I've got lots of time and plenty of electricity for this. | 04:56 |
systemdlete2 | Please, somebody. Talk me out of this. | 04:57 |
fsmithred | talk you out of installing everything? | 04:58 |
yanmaani | you'll get conflicts | 04:58 |
fsmithred | Don't install everything!!! | 04:58 |
yanmaani | Although I wonder what'd happen if you did sudo apt-get install -y * | 04:59 |
fsmithred | install what you need. If you want to keep it lean, exclude Recommends | 04:59 |
systemdlete2 | Why would libsnmp30, which I think has to do with email and email servers, depend on hardware sensors? | 05:00 |
yanmaani | just gives "you've held broken packages" error | 05:00 |
yanmaani | systemdlete2: look on the package info site | 05:00 |
systemdlete2 | And why am I suddenly getting withheld packages? Why, what did I do? I've been faithfully updating and upgrading, minding my own business. | 05:00 |
systemdlete2 | Is apt-get demanding a ransom I don't know about? | 05:01 |
systemdlete2 | If I broke something, I'll pay for it, I promise! | 05:01 |
systemdlete2 | apt full-upgrade seems to resolve whatever issues it had. It will remove packages that would cause a conflict, which apt upgrade should be doing in the first place. Unless one prefers broken configurations I guess. | 05:12 |
systemdlete2 | A package management system is useless if it does not ALWAYS enforce consistency across package dependencies. | 05:13 |
systemdlete2 | yanmaani: Do you mean https://pkginfo.devuan.org/stage/beowulf/beowulf/libsnmp30_5.7.3+dfsg-5.html ? | 05:23 |
systemdlete2 | Oh. My bad. | 05:24 |
systemdlete2 | snmp, not smtp. | 05:24 |
systemdlete2 | Still, I don't get why upgrading packages would not, by default, modify or remove or add packages as needed. It even tells you before you agree that it will be doing those things, so what is the difference? | 05:25 |
systemdlete2 | Is there any reason not to use "full-upgrade" rather than "upgrade" -- will it introduce problems that plain "upgrade" will not? | 05:25 |
yanmaani | systemdlete2: I recall being recommended to do upgrade first, then dist-upgrade | 05:26 |
systemdlete2 | dist-upgrade does something else entirely. | 05:26 |
systemdlete2 | Not what I want. | 05:26 |
yanmaani | what do you want? | 05:27 |
systemdlete2 | full-upgrade, it seems. | 05:27 |
systemdlete2 | doesn't dist-upgrade take you to the next release level or something? | 05:27 |
systemdlete2 | like ascii -> beowulf, e.g. | 05:27 |
systemdlete2 | (I DON'T want that!) | 05:28 |
rrq | no dist-upgrade stays with the sources.list points | 06:37 |
GNUmoon | Still trying to get thunderbolt 3 dock working with devuan. Just used a live image of Pop!OS and the secondary screen came up straight away... | 13:54 |
GNUmoon | Uses older kernel 5.4. | 13:54 |
GNUmoon | Logs a bit harder to read, given systemd...as no Xorg.0.log or messages. | 13:56 |
GNUmoon | only obvious difference is bolt version on Pop!OS is 0.8 | 13:57 |
GNUmoon | Devuan is at 0.7. | 13:57 |
djph | that might do it | 13:57 |
GNUmoon | I've added backports and experimental to sources.list, but nothing newer. | 13:57 |
GNUmoon | Do I need to compile from sources? | 13:57 |
GNUmoon | djph: yeah, its the most obvious thing to fix next :) | 13:58 |
GNUmoon | I hope it works. | 13:58 |
djph | me too :) | 13:58 |
fsmithred | 0.9-1 is in chimaera/ceres. | 14:03 |
GNUmoon | fsmithred: so do I just add "deb http://deb.devuan.org/merged chimaera main" to the source.list? | 14:07 |
fsmithred | that would be risky | 14:07 |
fsmithred | maybe download the package from packages.debian.org and install with dpkg -i | 14:07 |
fsmithred | or if you want to backport it, pull the source package with apt | 14:08 |
fsmithred | Depends: libc6 (>= 2.28), libglib2.0-0 (>= 2.56), libpolkit-gobject-1-0 (>= 0.99), libudev1 (>= 183) | 14:08 |
fsmithred | good chance it'll work in beowulf | 14:08 |
GNUmoon | fsmithred: thanks, I'll try the dpkg -i approach first. | 14:09 |
GNUmoon | Is there a guide for "...backport it, pull source package with apt"? | 14:10 |
fsmithred | yeah, I think I can find it | 14:11 |
fsmithred | https://wiki.debian.org/Packaging/SourcePackage | 14:13 |
GNUmoon | fsmithred: Thanks, I'll check it out. | 14:17 |
ham5urg | Has anyone worked with jq? How do I escape a json list to a single line for use in bash? | 14:56 |
r3boot | jq -c | 14:59 |
ham5urg | echo `jq -rc .packages <<< $SERVER` gets me ["pkg1","pkg2"...] | 15:01 |
ham5urg | I would need pkg1 pkg2 ... without escapes and commas. | 15:02 |
zatumil | <<<'["pkg1", "pkg2"]' jq -rc '.[]' - | 15:07 |
ham5urg | Thanks | 15:15 |
SuaveDandy | Does anyone know what "blockdevice" means in "od -j containersize - blocksize" on this ArchWiki page: https://tinyurl.com/dmcrypt-wipe ? | 17:37 |
SuaveDandy | Yes, I'm wiping with dm-crypt. | 17:38 |
MinceR | whatever the block device you tried to wipe was | 17:39 |
SuaveDandy | *blocksize | 17:39 |
SuaveDandy | Accidentally wrote "blockdevice." | 17:40 |
fsmithred | SuaveDandy, there's more to exclude from my script if you want to use it. (it tries to mount stuff) | 17:40 |
SuaveDandy | Ohhhhhhhh… | 17:41 |
MinceR | dunno | 17:41 |
SuaveDandy | Good that I haven't started the installation yet. | 17:41 |
SuaveDandy | Do you have the lines stored at paste.debian.org ? | 17:42 |
fsmithred | no | 17:42 |
fsmithred | Remove lines 1113-1177 | 17:42 |
fsmithred | uh, that only works if you didn't remove the other lines | 17:43 |
SuaveDandy | Alright. I'll add that to my note. | 17:43 |
SuaveDandy | Oh. | 17:43 |
SuaveDandy | Here's the lines you gave me. https://paste.debian.net/1163235/ | 17:43 |
fsmithred | from this comment: # make mount point, format, adjust reserve and mount | 17:43 |
SuaveDandy | I saved it in a note-taking app. | 17:44 |
fsmithred | to these lines: | 17:44 |
fsmithred | mount $boot_dev /target_boot | 17:44 |
fsmithred | sep_boot_opt="--exclude=/boot/*" | 17:44 |
fsmithred | fi | 17:44 |
fsmithred | also | 17:44 |
fsmithred | you told me two volumes that are mounted at /mnt. It doesn't work that way. | 17:45 |
fsmithred | I assume you have to install everything to one partition and zfs figures it out. Otherwise, I have no idea what it means. | 17:45 |
fsmithred | or you move stuff manually afterward | 17:46 |
SuaveDandy | ZFS uses two partitions as two pools. | 17:46 |
fsmithred | if you mount two devices at the same mountpoint, you only see the second one. I can't get my head around what you're saying. | 17:46 |
SuaveDandy | fsmithred: Here's the whole documentation. https://tinyurl.com/zfs-on-root | 17:48 |
fsmithred | oh, you might need to add -X to the rsync command. | 17:50 |
fsmithred | I see xattr mentioned | 17:50 |
SuaveDandy | What about the lines to delete? | 17:52 |
SuaveDandy | Do I delete the lines you sent me before or the lines you sent me now? | 17:52 |
fsmithred | both | 17:53 |
fsmithred | you said you created the mountpoints and mounted everything manually, right? | 17:53 |
fsmithred | that's what the second group of lines does, but not the way you want. | 17:54 |
fsmithred | and just so you know, I'm not able to read that page and then tell you what you're supposed to do. | 17:55 |
fsmithred | I would have to actually do it myself. | 17:55 |
SuaveDandy | Alright. I'mma try using debootstrap again. This time not using --include and --exclude | 17:56 |
fsmithred | that's probably easier | 17:57 |
SuaveDandy | This is my last chanse. After that I give up. | 17:57 |
fsmithred | less chance of having to re-do it | 17:57 |
SuaveDandy | Such a shame ZFS uses CDDL. | 17:58 |
SuaveDandy | We have to go through all of this because of that | 17:58 |
SuaveDandy | Oracle are big meanies. | 17:58 |
SuaveDandy | Although I heard it's because the OpenZFS team can't find some devs to sign for the change of license. | 17:59 |
SuaveDandy | Some gone, some dead. | 17:59 |
nemo | does anyone have recommendations for fixing the AMD rdrand bug on Devuan Beowulf? | 18:29 |
nemo | I'm thinking I probably should be asking this in #debian | 18:29 |
nemo | but figured I'd check in the channel I'm idling in first | 18:30 |
nemo | followed some instructions months ago, that did not fix it | 18:30 |
nemo | thankfully it is not as serious on devuan as on systemd distros. just kind of annoying in qt apps | 18:30 |
DHE | so I run a different distro but had an rdrand issue with a ryzen 3000 series chip. openssh and openssl apps refused to work. solution was a mobo bios update | 18:36 |
nemo | DHE: wut | 18:37 |
nemo | DHE: openssh and openssl use rdrand exclusively??? | 18:37 |
nemo | DHE: hm. maybe in Windows. I can totally see that there | 18:37 |
nemo | where proper OS support for random numbers is a recent thing | 18:37 |
nemo | but wait you said distro | 18:37 |
nemo | DHE: which app | 18:37 |
DHE | I mean not devuan | 18:38 |
SuaveDandy | DHE: You run a different distro? What do you use Devuan for then? | 18:40 |
DHE | right now I'm testing it in a VM for a new system | 18:40 |
SuaveDandy | Ah. | 18:52 |
SuaveDandy | If you've tested Debian, you've kinda tested Devuan. | 18:52 |
SuaveDandy | I didn't even bother testing in VM. It's not like Arch or NixOS. | 18:53 |
DHE | maybe but I hate systemd with a burning passion. in a "pour gasoline and burn it" sort of way. | 18:59 |
nemo | DHE: can you tell me which application uses rdrand as an exclusive source of randomness? | 19:48 |
nemo | DHE: or... could it be that this dubious distro uses rdrand exclusively as a kernel entropy source? | 19:48 |
yanmaani | windows has proper os rng | 20:06 |
yanmaani | vm rng is sometimes fucked | 20:07 |
SuaveDandy | mason: How hard is it to maintain ZFS compared to Ext4? | 20:07 |
yanmaani | in general, not windows | 20:07 |
mason | SuaveDandy: Fairly easy regardless of whether you're using DKMS or custom packages. | 20:08 |
mason | SuaveDandy: You need to make sure you schedule scrubs, so that's a little more effort. | 20:08 |
mason | Without scrubs, you risk things going wrong and not getting caught in time to fix. | 20:09 |
SuaveDandy | And the ZFS guys said noobs shouldn't use such an expert system and should use Ext4 instead. | 20:09 |
SuaveDandy | So I thought. | 20:10 |
fsmithred | 20 years of linux here and still using ext* | 20:13 |
SuaveDandy | I don't know if I should go forward. I mean, will I have enough time to learn ZFS, set up my whole system with window manager and work? | 20:15 |
SuaveDandy | I don't even know. | 20:15 |
SuaveDandy | But I mean, on Windows I always had a long acclimatization period after installing anyway. | 20:16 |
SuaveDandy | And still didn't have the things my way like the keyboard layout. | 20:16 |
SuaveDandy | Or the window layout. Or the popup terminal. | 20:16 |
fsmithred | takes me a few weeks to get a new system all set up | 20:16 |
SuaveDandy | Do you think it's a better idea to learn ZFS in VM or I can learn it on-the-fly? | 20:19 |
fsmithred | I don't know it, so I can't estimate. | 20:19 |
SuaveDandy | mason? | 20:20 |
yanmaani | how can it take weeks to set up a file system | 20:20 |
yanmaani | ext4 takes minutes | 20:20 |
mason | SuaveDandy: I'd use it live. That's how I did it anyway, to learn. | 20:21 |
fsmithred | not to set up a file system | 20:21 |
fsmithred | weeks to get my installation the way I want it | 20:21 |
yanmaani | I've had this crazy idea | 20:21 |
mason | SuaveDandy: The tricky bits are getting comfortable sending/receiving snapshots for incremental backups, etc. | 20:21 |
yanmaani | you make a "DE" | 20:21 |
SuaveDandy | Live as in on a live CD? | 20:21 |
yanmaani | which is just "what normal people want for their system" | 20:21 |
fsmithred | initial install of the OS only takes me 10 minutes from a live iso | 20:21 |
yanmaani | where normal people == you | 20:21 |
fsmithred | yeah, usually DE | 20:21 |
mason | SuaveDandy: And since ext does nothing like that, that's really like learning different tools, unrelated to the files sitting there and being accessed | 20:21 |
yanmaani | so, mpv for media player, FF ESR for browser, that sort of thing. | 20:22 |
SuaveDandy | mason, what do you mean by "live?" | 20:22 |
fsmithred | I think he means "on hardware and in use" | 20:23 |
mason | that | 20:23 |
fsmithred | not a test run | 20:23 |
fsmithred | like getting thrown in the water to learn how to swim | 20:24 |
fsmithred | SuaveDandy, did you try the deboostrap install on zfs? | 20:25 |
SuaveDandy | Not yet. | 20:25 |
SuaveDandy | Was discussing stuff. | 20:26 |
SuaveDandy | Gimme a sec. | 20:26 |
fsmithred | how long did it take to set up the zfs stuff? | 20:26 |
fsmithred | or better, how long would it take to do it again, now that you know how? | 20:26 |
SuaveDandy | Well, heh. A couple of half an hour. | 20:31 |
SuaveDandy | I'mma speed up, guys. | 20:40 |
SuaveDandy | Focus mode: on | 20:40 |
user_ | Devuan mentioned https://www.theregister.com/2020/09/10/debian_project_address/ | 20:55 |
* user_ chuckles a bit, muahaha, not enough people to perfect total Poetteringization? or not enough people to prevent too much breakage by that prolific man. | 20:56 | |
user_ | hehe comments start with a very much #devuan close question >:-) https://forums.theregister.com/forum/all/2020/09/10/debian_project_address/ | 20:58 |
* user_ is done with noise, people can read if they want to. The links. | 20:58 | |
* user_ chuckles as usual reading the comments, thereg comments are worth as much as the article(s) imo | 21:00 | |
golinux | Please take it to #debianfork | 21:01 |
user_ | np, done | 21:02 |
SuaveDandy | So. We have configs and user data backed up to the cloud. And the whole system backed up to the external drive, say, once in 6 months. The hardware failure is unlikely to happen. Does that mean that ZFS snapshots are redundant? | 22:29 |
SuaveDandy | mason | 22:29 |
SuaveDandy | I'm not going to do anything stupid. | 22:30 |
SuaveDandy | Although I may mess the software up… | 22:31 |
SuaveDandy | Makes sense. | 22:31 |
mason | Snapshots are useful as point-in-time references on top of their use in doing incremental back-ups. | 22:32 |
SuaveDandy | Something like restore points? | 22:32 |
mason | Yes. | 22:32 |
SuaveDandy | I just thought that Debian is super reliable. | 22:33 |
gnarface | SuaveDandy: Debian is super reliable. You, however, by comparison, are not. | 22:34 |
SuaveDandy | Heh. | 22:34 |
SuaveDandy | Well, I'm only human. | 22:34 |
gnarface | SuaveDandy: sometimes you only realize you made a mistake months later. that's when it's good to have historical snapshots to refer to | 22:34 |
gnarface | SuaveDandy: we are all only human. well, most of us are. i might not be, but i'm no less error prone for it. | 22:35 |
gordonDrogon | when I was being paid to sysadmin, users deleting files was very common. less common was "can you get that version of the file I had 3 weeks ago", but it did happen. | 22:36 |
gnarface | ransomware attacks and mysql database corruption also come to mind | 22:37 |
specing | I have hourly snapshots and they had saved me dozens of times already | 22:44 |
specing | and half of those dozens of times were when messing with git | 22:44 |
specing | and nuking my working checkout by accident | 22:45 |
specing | the snapshots are kind of like revision control system control system | 22:45 |
mason | Ooh, I may have identified my startup race. At least, this matches what I see: https://linux.debian.bugs.dist.narkive.com/a99qRmYJ/bug-912237-etc-init-d-rpcbind-stat-not-found | 23:28 |
mason | Looks like I've been bitten by a UsrMerge bug. | 23:30 |
mason | This could be another one for us to fork. | 23:31 |
mason | I'll see if there's a Debian-side bug. | 23:31 |
mason | bbiab, rebooting, testing | 23:31 |
unixbsd | is XFS on devuan more stable than EXT4? | 23:43 |
gnarface | it is more stable period. | 23:48 |
gnarface | but that doesn't mean it will be the best choice in every situation (for example, Valve's Steam client for Linux has had a history of obscure patch failures on anything other than ext4) | 23:49 |
gnarface | also, even though generally everything is faster with XFS, deletions are slower, so there may be very different performance implications depending on the work load | 23:50 |
gnarface | i would just suggest to try it | 23:51 |
gnarface | if you want faster launch times and more read throughput, XFS is a shoe-in unless you only have a single-core system (XFS is also the only multithreaded filesystem) | 23:52 |
gnarface | on the other hand, if you want a build server for kernels, and you do a lot of "rm -rf /usr/src/*" and the like, the delay time might get annoying | 23:55 |
gnarface | not like, cripplingly slow, but if you are used to all deletions being basically instantaneous you might get frustrated having to wait 30s to delete 2GB of small files | 23:57 |
Wonka | Back Then[tm], when I used XFS, after crashes all files that had been open for writing during the crash were completely zeroed. Didn't happen with ext3. (ext4 didn't exist yet) | 23:57 |
gnarface | Wonka: it would put the unfinished file chunks in /lost+found/ and you just had to identify them and you could usually manually put them back in place. i haven't seen it have that problem for several years now. i think they changed default mount options and did some internal tuning. | 23:58 |
gnarface | Wonka: in the mean time, the ext4 team can't even keep slightly outdated versions of e2fsprogs from unraveling the whole filesystem | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!