Search results

Now to figure out how to get that space reclaimed on the backup server that I zfs send to... (without needing a full re-transfer)..

Yeah that did it. (On a test box I made a copy of the zvol to not mess with production. Will fix on production shortly.) Spose have to be careful to ensure discard is on (and wonder why it's not a default on..).

I guess a secondary question is can I turn on discard now (and reboot the VM) and then run e2defrag for eg?

Looked around for an answer to this but didnt quite find (except fstrim). I'm thinking that a VM's ext4fs ate a lot of disk, files subsequently deleted and the zvol remains large, despite being thin provisioned. I didnt have discard on because of oversight. Id like to get the space back...

AFAIR Apache and MariaDB also need nesting these days...

the weird fuse mount of /etc/pve causes issues. i ran out of disk on / now its unmodifyable for eg. I think I need to reboot :(

And where do these templates get downloaded from? Ie url on the internet to retrieve manually for eg. [SOLVED]: https://images.linuxcontainers.org/images/

Yes I use screen all the time, and I have one on the host, with one window per container. I wanted to save that one screen window login cuz I was already maxed at 50 screen windows. Yes, easy to escape the ssh, I want to escape the PCT ENTER itself, not the ssh.

Im ssh'd in to the host, course I can ssh in again and kill the session, was just wondering if there's a cleaner escape sequence. Seems there is supposed to be, but it does not seem to work.

Can't exit from a wedged container. It's frozen, dead, etc. I want to escape the PCT enter and diagnose from the host without having to open a new shell etc. Should be possible regardless of workarounds.

long time user but somehow stuck on a noob problem - how to escape PCT enter? had a container wedge, wanted my shell back on the host. this does NOT seem to be in play (no, Im not using it in screen with meta=^a) from pct(1): --escape \^?[a-z] (default = ^a) Escape sequence prefix. For...

D Dont need to reboot if you do the echo stuff as above. But echo wont survive a reboot, so have to update-initramfs to make permanent with the changes to /etc/modprobe.d/zfs.conf

Same error, different problem: somehow the host's /dev/pts got remounted ro, only when starting a CT did it popup til later I tried creating a new screen window 'read-only filesystem'. Chased it down to /dev/pts being ro, did a mount -o remount etc. So there are many causes to this error...

Had same issue, did a systemctl restart pve-cluster and it cleared up.

Aha rereading your answer it seems you did answer, apologies, but it wasnt immediately clear. "As pve-cluster doesn't depend on a specific version this is fine" was the key part.

bump. This question was not answered. Seems key since many utilities are using fuse3 and dont want to pull the fuse package rug out from under pve-cluster et al.

Yep breaks everything. This instruction is wrong and dangerous. (PVE 7.1 here)

(ie: NOT SOLVED)

so then how to reduce impact?

Install 7.x and upgrade. However I think 7.x removed the HOLD-ALT-TO-HIGHLIGHT-KEY-TO-CHOOSE-BUTTON for keyboard only -- my servers dont have a lot of USB (which are user periferals..) - 2 USB slots: 1 for USB key and one for keyboard. Should be able to install with only keyboard! These are...