| === lotuspsychje__ is now known as lotuspsychje | ||
| lordievader | Good morning | 07:25 |
|---|---|---|
| brektyme | Hi, I'm getting a libc segfault when pxebooting an install during the disk detection. Does anyone have any ideas? | 13:41 |
| blackflow | brektyme: hard to tell unless you trace something somwhere to see where it faults. otoh, is the booting OS correct for the hardware and architecture? | 13:43 |
| brektyme | yes | 13:43 |
| brektyme | I'm using the kernel and initrd from the mirrors, both amd64 | 13:44 |
| brektyme | which is correct for the hardware | 13:44 |
| brektyme | the kernel and initrd from here: http://archive.ubuntu.com/ubuntu/dists/xenial/main/installer-amd64/current/images/netboot/ubuntu-installer/amd64/ | 13:48 |
| brektyme | using the kernel and initrd from here gets past the segfault, but was causing issues with installation later around not being able to find kernel modules: http://archive.ubuntu.com/ubuntu/dists/xenial-updates/main/installer-amd64/current/images/netboot/ubuntu-installer/amd64/ | 13:49 |
| brektyme | although just testing now with the newer kernel and initrd it had no issues with installation. | 13:51 |
| DammitJim | has anyone set up drbd on ubuntu servers and use pacemaker to not only ensure who is the master, but also trigger the start of a service? | 15:05 |
| sdeziel | DammitJim: drbd works really well (tested on every LTS from 10.04 to 16.04 so far) but not with pacemaker | 15:23 |
| DammitJim | sdeziel, how are you using it w/o pacemaker? | 15:24 |
| DammitJim | thanks for your info, btw | 15:24 |
| DammitJim | man, a lot of people don't like glusterfs and recommend ceph besides drbd | 15:24 |
| sdeziel | DammitJim: I'm using it through ganeti (cluster manager) that automates the creation of DRBD volumes that get assigned to QEMU VMs | 15:24 |
| sdeziel | DRBD is kind of old school | 15:24 |
| DammitJim | holy shnikies! | 15:24 |
| DammitJim | but drbd is pretty solid, right? | 15:25 |
| sdeziel | oh yeah | 15:25 |
| sdeziel | we've been live migrating VMs since many years | 15:25 |
| sdeziel | no DRBD issue to report | 15:25 |
| DammitJim | how do you share the drbd file system to be accessed by clients? | 15:26 |
| DammitJim | like is this something where I can have a process in one of the drbd servers (or on all of them) where the process writes locally to the file system? | 15:27 |
| sdeziel | the clients are QEMU processes so all they see is a regular block device like you'd get from a LVM slice | 15:27 |
| sdeziel | DammitJim: yes you can have that | 15:27 |
| DammitJim | oh, interesting... they see a block device | 15:27 |
| DammitJim | so, say 2 drbd servers | 15:27 |
| sdeziel | DammitJim: you can see DRBD as RAID1 over the network | 15:27 |
| sdeziel | read are served by the local copy if possible with a fallback to the peer | 15:28 |
| DammitJim | then the qemu process runs on another server, but can see the RAID1 over the network to access the block device for a new VM? | 15:28 |
| sdeziel | we have say /dev/drbd1 on node01 and node02. A given VM running off of node01 uses /dev/drbd1. During a live migration, the QEMU process is "moved" to node02 which already has the same disk content at /dev/drbd1 so only QEMU's memory needs copying | 15:29 |
| DammitJim | yikes | 15:30 |
| sdeziel | as far as QEMU is concerned it writes to a local disk, end of story | 15:30 |
| sdeziel | DRBD does the magic behind the scene | 15:30 |
| DammitJim | that's insane | 15:31 |
| sdeziel | insane in the good way? | 15:31 |
| DammitJim | I think I understand it, but my brain might still be behind processing this | 15:31 |
| DammitJim | insanely good, yes | 15:31 |
| sdeziel | yes it is | 15:31 |
| DammitJim | do you remember if you had to zero out the hard drive before configuring drbd? | 15:36 |
| DammitJim | I don't know why I'm running into issues with /dev/sdb1 where it doesn't allow me to add it | 15:36 |
| sdeziel | DammitJim: I'd refer you to https://docs.linbit.com/docs/users-guide-8.4/#ch-configure as it's been an eternity that I manually configured a DRBD device manually (ganeti does it all for us) | 15:39 |
| sdeziel | DammitJim: unless you have unrelated partitions on sdb, I would recommend to either directly feed /dev/sdb to DRBD or put LVM/etc under it | 15:41 |
| sdeziel | partitions make it hard to resize later down the road | 15:41 |
| DammitJim | gosh, LVM might make sense in case that someone decides we need more storage | 15:41 |
| DammitJim | so, I can even do this with LVM?!!! | 15:41 |
| sdeziel | that's what we use | 15:42 |
| DammitJim | gosh, I have a lot of work ahead of me | 15:42 |
| sdeziel | LVs under and DRBD over | 15:42 |
| DammitJim | I need to take baby steps | 15:42 |
| sdeziel | I think the LVM way is best even for experimenting | 15:42 |
| DammitJim | totally separate question (I need to get to other important stuff)... any recommendations to join a Windows domain? | 15:43 |
| DammitJim | I've run into so many issues with samba/kerberos, etc... even though I have a somewhat working solution | 15:44 |
| DammitJim | the systems seem to break that funcionality when upgrading from Ubuntu 16 to 18 | 15:44 |
| fearnothing | hola | 22:43 |
| fearnothing | having some DNS problems | 22:43 |
| fearnothing | dig @8.8.8.8 www.google.com - works fine | 22:43 |
| fearnothing | dig @192.168.x.x www.google.com - no packets get sent to 192.168.x.x | 22:44 |
| fearnothing | wtf | 22:44 |
| sarnold | is that address routable? check with ip route get | 22:45 |
| fearnothing | it's on the same subnet | 22:45 |
| fearnothing | fml | 22:47 |
| fearnothing | nevermind | 22:47 |
| fearnothing | found the problem | 22:47 |
| fearnothing | I'm a dingus | 22:48 |
| sarnold | ooh, what was it? :) I always like learning new ways to be a dingus ;) | 22:50 |
| teward | fearnothing: wrong IP? DNS server not working right? Nonexistent DNS server at address? | 22:51 |
| teward | we are curious | 22:51 |
| teward | ohai sarnold there be work for you :p | 22:51 |
| fearnothing | re-IPed my DNS server 3 months ago didn't I | 22:52 |
| teward | hah | 22:52 |
| fearnothing | updated all my other hosts' resolv.conf files but forgot this one | 22:52 |
| teward | ERR: Wrong IP in DNS nameservers list. | 22:52 |
| teward | I've done that too :P | 22:52 |
| fearnothing | right, time to do-release-upgrade because I've been a lazy mofo | 22:53 |
| sarnold | fearnothing: oooh that's a good one :) | 22:53 |
| fearnothing | (?!.*regex)I've got(?:\s(?<problem>[^\s]+)){99} | 22:55 |
| fearnothing | yeah for some reason I had used a 32-bit image for my bind server, and that meant I couldn't install something else I needed | 22:56 |
| fearnothing | I mean wtf it was only installed 2 years ago where did I even get a 32 bit image | 22:57 |
| sarnold | "a dns server probably doesn't need much address space anyway" | 22:57 |
| fearnothing | that's ancient history man | 22:57 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!