=== vrubiolo1 is now known as vrubiolo | ||
otubo | smoser, hey, don't see anything regarding LVM on cloud-utils-growpart, would have plans to do it? I have a couple of issues with customers using RHEL requesting it. | 11:40 |
---|---|---|
Odd_Bloke | We just landed a fix for an issue that affects groovy (thanks rharper!), so I'm going to upload a new cloud-init to groovy. | 13:09 |
meena | what's groovy? | 13:16 |
rharper | Odd_Bloke: nice! | 13:28 |
rharper | meena: new ubuntu development release name Groovy Gorilla | 13:28 |
meena | 20.10? | 13:29 |
meena | yes | 13:29 |
Odd_Bloke | rharper: If you have a minute to validate: https://github.com/canonical/cloud-init/pull/486 | 13:32 |
smoser | https://bugs.launchpad.net/cloud-init/+bug/1799953 | 13:43 |
ubot5 | Ubuntu bug 1799953 in cloud-utils "growpart does not work for lvm" [Medium,Triaged] | 13:43 |
smoser | otubo: ^ . I have no plans. but ^ is the bug that was created for the issue. | 13:43 |
meena | works for zfs | 13:48 |
meena | but then again, zfs is lvm + $fs in one | 13:48 |
otubo | smoser, thanks | 13:49 |
otubo | Odd_Bloke, do you guys plans to migrate issues from launchpad to github as well in the future? | 13:49 |
Odd_Bloke | otubo: Nope, we have to use Launchpad for Ubuntu issues/process regardless, so migrating "upstream" issues to GitHub would just leave us handling issues across two platforms. | 13:51 |
smoser | otubo: https://bugs.launchpad.net/cloud-init/+bug/1799953 | 14:25 |
ubot5 | Ubuntu bug 1799953 in cloud-utils "growpart does not work for lvm" [Medium,Triaged] | 14:25 |
smoser | comment added there. it seems pretty easy. if you wanted to try | 14:26 |
meena | smoser: the only issue i see is using the device uuid, or some other link to it, instead of the device path | 14:43 |
smoser | meena: ? | 15:10 |
smoser | how so. | 15:10 |
meena | i dunno, hmm, maybe that just means more steps in the resolution of the name / path | 15:11 |
smoser | you mea if the pvcreate had been done with a different path ? hm... | 15:14 |
smoser | i dont know how names for pvs are done. | 15:14 |
smoser | so yeah, th3e check that is uggested (pvs | grep) could be an issue. and result in not running pvresize. | 15:16 |
smoser | but it seems the other pv commands when given a block device path probably open it, read the PV UUID and use *that* | 15:16 |
smoser | just playing... i did a 'pvcreate' with the by-partuuid path: | 15:19 |
smoser | pvcreate /dev/disk/by-partuuid/2079bb1d-d7ca-4b4f-a1d1-a62c06b7c570 | 15:19 |
smoser | and then 'pvs' will show the pv with the 'loop5p1' name. I'm guessing that it just does effectively 'readlink -f' on whatever input for that. | 15:20 |
smoser | thanks alot meena | 15:25 |
smoser | now i'm back to being depressed about linux storage | 15:26 |
rharper | s/linux storage/linux-lvm2 | 15:54 |
meena | i wish zfs would work somehow across computers,so you could build cloud storage | 16:17 |
meena | but chances are that then it simply wouldn't work | 16:17 |
smoser | rharper: wrt sucking, i mean that wrt almost all of linux storage. its not limited to lvm. | 16:33 |
meena | imagine if ceph was easy! | 16:39 |
smoser | it woudl still suck. | 18:00 |
smoser | sorry.... rharper and I submitted a talk to open source summit, but it didn't get accepted. | 18:01 |
smoser | things we learned and got burnedon in curtin and other places. | 18:02 |
meena | smoser: about storage on Linux? | 18:22 |
meena | did you still write it down? | 18:22 |
smoser | https://hackmd.io/eMOklfSvS6aFCcmu_1h79Q | 18:26 |
meena | wow, this is literally the worst web page i've visited all day: http://www.golismero.com/ | 18:28 |
meena | smoser: cool! i … didn't expect that, but wow, cool! | 18:28 |
meena | uuids are not unique when you snapshot and re-attach. this goes for partition uuids, and fs uuids. What is a good way to find the right path to this thing. fstab ⬅️ wat | 18:44 |
rharper | smoser: heh; lvm2 has a special place on my crap list ... | 18:44 |
rharper | meena: https://wiki.ubuntu.com/FSTAB | 18:44 |
rharper | consolidated list of best persistent links; we generally prefer to avoid FS_UUID if possible, as it;'s mutable; but in the absence of any other one, we use that | 18:46 |
smoser | all that magic (like LABEL= or UUID=) it all works most of the time. | 18:46 |
smoser | and when a human is typing stuff, the times ti doesn't work, they just live with and reboot or whatever. they say "oh well, that doesn't work because x y z" | 18:46 |
meena | i can copy/paste, or i can use human readable / writable paths. | 18:50 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!