| znf | Does anyone know of a tool like https://regresshion.io/api ? (preferably self-hosted) | 00:28 |
|---|---|---|
| === chris14_ is now known as chris14 | ||
| Odd_Bloke | Does anyone know when we can expect to see a noble release cloud image with the OpenSSH fix? (I see there's a daily containing it: http://cloud-images.ubuntu.com/noble/20240702/) | 16:04 |
| tuxick | cloud-init autoinstall on noble-server-cloudimg-amd64.img gives me "Skipping autoinstall module. Expected one of the Ubuntu installer snap packages etc" | 16:37 |
| tuxick | why? | 16:38 |
| tuxick | goal is to deploy on proxmox, passing --cicustom | 16:39 |
| === sdeziel_ is now known as sdeziel | ||
| dbungert | tuxick: autoinstall is not used with cloud images. You can use cloud-init directly in that case. | 18:34 |
| dak_ | I have an ZFS storage server (Dell R630) running ubuntu 22.04.4 which is sharing these ZFS datasets over NFS. This server has suddenly been having horrible performance issues, which seem to be getting worse over time. For instance, writing a 1G image file takes longer than it should and is written at 1 to 2 MB/s. It should be in the 200 of MB/s at least. I have tried rebooting the server, upgrading firmware, Memtests, running an older kernal, remounting | 21:53 |
| dak_ | shares, nothing has worked so far. I have even mounted one of the NFS shares locally on the NFS server (localhost) itself and I'm still having horrible issues with performance when writing/reading from the share. Suggesting this is something wrong with NFS itself and not a networking related issue. But what? Thoughts on what I might try next? | 21:53 |
| sarnold | how full are they? how much free space? dmesg? zpool status? smartctl -a ? fio to the drives? fio to the dataset? | 21:57 |
| dak_ | Not full at all TBs free, pool is healthy, performance is great when reading and writing to a dataset directly. I can write a 1GB file at 1.2GB/s without issue. But 1-2 MB/s through NFS. Which is far far below what we typically get. | 22:00 |
| dak_ | I didn't see anything in dmesg when I was poking around, but I can look again. | 22:01 |
| dak_ | I can also tell you that nfsstat tells me that about 10-12% of RPC stat calls are logged as badcalls, specifically badfmt. At the moment, the stats are Calls: 61325 badcalls: 7628, badfmt: 7628, badauth:0, badclnt:0 | 22:08 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!