| === TheHonorableKitt is now known as ThKitten | ||
| cpaelzer | good morning | 05:58 |
|---|---|---|
| lordievader | Good morning | 07:14 |
| === Bodenhaltung_ is now known as Bodenhaltung | ||
| ahasenack | good morning | 10:31 |
| baffle | Hi, I am trying to get frr to start on boot on 18.04, and the unit-file has Wants/Before=network-online.target .. But network-online.target is never reached, status says "inactive (dead)".. network-online.target has depends on systemd-networkd-wait-online.service (I use netplan for interface config), but that is also "inactive (dead)". It again depends on systemd-networkd.service, wich is "active | 14:39 |
| baffle | (running)" and has no errors, and have configured interfaces.. Any tips? I'm getting really frustrated. If I do "systemctl --type=target" network-online.target is not listed.. | 14:39 |
| baffle | If I switch back to using /etc/network/interfaces, network-online.target has other dependencies, but still doesn't trigger. | 14:41 |
| TJ- | baffle: if you've got a Wants= surely you need an After= not a Before= | 14:59 |
| baffle | TJ-: Sorry. /lib/systemd/system/frr.service has [Unit] After=networking.service, [Install] WantedBy=network-online.target. | 15:17 |
| baffle | TJ-: If I do systemctl show frr.service it says "Before=shutdown.target network-online.target" and "After=basic.target system.slice networking.service systemd-journald.socket sysinit.target". | 15:19 |
| TJ- | baffle: ahhh... maybe I've got confused again over 'Wants=' - I seem to recall someone in #systemd telling me it is like an inverse dependency, as in this unit becomes Want-ed by the named unit | 15:34 |
| theGoat | so i would like to centrally manage the authentication of my linux VMs, without using AD. was wondering what my options were? | 15:56 |
| cyphermox | LDAP | 16:04 |
| cyphermox | also NIS, and probably something I forgot, too | 16:05 |
| leftyfb | Anyone know of some documentation on deploying a disk image (from dd maybe) using PXE? (Not the fog project) | 17:41 |
| lordcirth | leftyfb, you probably want clonezilla | 17:46 |
| leftyfb | nope | 17:47 |
| leftyfb | clonezilla won't use a dd image | 17:47 |
| leftyfb | we want to create images in a CI system and then deploy them using PXE | 17:47 |
| lordcirth | That's surprising. Have you considered not using dd, then? It is pretty inefficient | 17:48 |
| leftyfb | possibly | 17:49 |
| lordcirth | Also, the clonezilla page says you can use dd images | 17:50 |
| lordcirth | "Based on Partclone (default), Partimage (optional), ntfsclone (optional), or dd to image or clone a partition." | 17:50 |
| teward | leftyfb: you can use advanced clonezilla server / client modes | 17:52 |
| teward | and make it use dd | 17:53 |
| teward | it's slow (and I mean SLOW) but | 17:53 |
| leftyfb | define slow | 17:53 |
| leftyfb | slower than a d-i install via pxe? | 17:53 |
| teward | no i mean the dd process :P | 17:53 |
| lordcirth | probably depends on bandwidth vs cpu | 17:53 |
| teward | ^ this | 17:53 |
| lordcirth | d-i is bottlenecked by dpkg unzip, I think | 17:54 |
| TJ- | leftyfb: what have you tried? are you using syslinux? | 17:54 |
| leftyfb | I must say, I was VERY impressed with FOG. It pull a 10G image from a 120G drive and compressed it down to about 600M, all within a minute or 2. Takes the same amount of time to deploy it back to the same drive | 17:54 |
| lordcirth | leftyfb, that's pretty cool | 17:54 |
| leftyfb | TJ-: yes, I'm using syslinux | 17:54 |
| lordcirth | So, why aren't you using fog, then? | 17:55 |
| TJ- | leftyfb: so you can do "LINUX memdisk" "INTRD disk.img" "APPEND raw" ? | 17:56 |
| TJ- | s/INTRD/INITRD/ | 17:56 |
| leftyfb | lordcirth: because it's way overkill for what we need, not sure about the importing of dd images from a separate CI system and there would be a ton of custom work to get it to use the system serial# from DMI to determine a hostname as opposed to MAC address. At the moment I can't even get it to change the hostname after deployment like it's supposed to | 17:56 |
| TJ- | leftyfb: oh, you wouldn't even need the APPEND for a HDD/partitioned image file > 4GB | 17:57 |
| leftyfb | TJ-: At the moment I'm only using syslinux to run a d-i install. Are you saying syslinux is capable of deploying a dd image on it's own? I'm not looking to run it live, I need it to dd the image back to the drive | 17:57 |
| TJ- | leftyfb: using memdisk, yes | 17:57 |
| TJ- | leftyfb: maybe this'll help: https://www.syslinux.org/wiki/index.php?title=MEMDISK#Hard_disk_images | 17:58 |
| leftyfb | TJ-: you sure this will dd the disk image to the physical drive and not just try to run it live in memory? | 17:59 |
| TJ- | leftyfb: if you want to write the image, you'd need to boot a minimal initrd.img that fetches the disk/img and writes it to the device | 17:59 |
| leftyfb | ah, see, right | 17:59 |
| leftyfb | That's what I'm looking to do, write the image | 17:59 |
| TJ- | leftyfb: it'd have to be 2 stage, you've got to have something that can do the fetching and writing | 18:00 |
| leftyfb | TJ-: is there documentation on writing such an initrd.img? Can it run a python script as well? | 18:00 |
| leftyfb | TJ-: what do you mean by 2 stage? As in chain loading initrd's or something? | 18:01 |
| TJ- | leftyfb: 2 stage as in 1) PXE boot a kernel and initrd 2) initrd fetches disk.img and writes it to disk, then sets things up to pivot to its rootfs and start its /sbin/init | 18:04 |
| leftyfb | ah right, for any post-install(HOSTNAME) right? | 18:04 |
| TJ- | leftyfb: casper is designed to do that but with an ISO image; you may be able to adapt those casper initialramfs scripts/config/directives to do it | 18:05 |
| TJ- | casper loads/fetches the ./casper/filesystemd.squashfs for example | 18:05 |
| leftyfb | The only thing I need to do post-image is change the hostname. I could do that just by mounting the partition and sed'ing the /etc/hostname | 18:05 |
| leftyfb | So maybe all I need is a custom initrd | 18:06 |
| TJ- | yes, it sounds like it | 18:07 |
| leftyfb | Can dd use an image it's getting live over wget or the like? It'll be in the 10's of GB's, not enough to stick in memory | 18:08 |
| lordcirth | leftyfb, dd will accept from a pipe. Get it there how you please | 18:08 |
| lordcirth | honestly you could probably netcat it | 18:08 |
| leftyfb | ok, that's what I figured. I've done it from tar/zip | 18:09 |
| lordcirth | netcat may not be a wise idea, though | 18:09 |
| leftyfb | looks like I have a ton of stuff to look into/learn | 18:09 |
| lordcirth | Don't we all | 18:09 |
| leftyfb | it's too bad FOG wasn't easier to play with. The imaging/deploying was so efficient | 18:10 |
| lordcirth | I've added FOG to my tech todo list | 18:10 |
| leftyfb | The install was a bit of a mess | 18:10 |
| lordcirth | I want to install OpenNebula on my home server, and put FOG in a VM on it | 18:11 |
| leftyfb | It's a well-meaning project and it's got the imaging stuff down nicely, but the UX is pretty bad | 18:11 |
| lordcirth | It's not a homelab if it's not needlessly complex for home use | 18:11 |
| leftyfb | lordcirth: https://photos.app.goo.gl/LmdPoSS9wP4VtCDt8 | 20:35 |
| leftyfb | my "home lab" | 20:36 |
| lordcirth | neat | 20:36 |
| leftyfb | lordcirth: have you seen my little pi rack? | 20:43 |
| lordcirth | no | 20:43 |
| leftyfb | https://photos.app.goo.gl/UBtwQK2mumzh4w7F9 | 20:47 |
| mybalzitch | oh shit, that's nicely organized and everything! | 20:48 |
| leftyfb | :) | 20:48 |
| mybalzitch | that 2nd photo is real pleasing to look at | 20:48 |
| leftyfb | all custom | 20:48 |
| lordcirth | Shiny | 20:50 |
| bjonnh | running docker in LXD : https://support.plesk.com/hc/en-us/articles/360012448554-Docker-updated-leads-to-the-failed-docker-service-Dependency-failed-for-Docker-Application-Container-Engine | 21:57 |
| bjonnh | I had to follow (some how, they have a lot of mistakes) that | 21:57 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!