mathstodon.xyz is one of the many independent Mastodon servers you can use to participate in the fediverse.
A Mastodon instance for maths people. We have LaTeX rendering in the web interface!

Server stats:

2.7K
active users

#zfs

20 posts20 participants0 posts today

I guess the way to go for a #FreeBSD storage solution which is HA is HAST (block level replication), CARP (VIP failover), #ZFS file system (with some scripts exporting/importing the file system from the HAST-replicated block devices on failover), a NFS server (which starts/stops together with the active ZFS replica), and a WireGuard tunnel to encrypt NFS traffic over the network. And some periodic ZFS snapshots and regular background scrubbing. Not sure if CARP can build a quorum with 3 nodes?

So many comments about not wearing a hat and glasses in the last one. Everything from losing my nerd credibility to thinking that it’s an AI avatar. Thanks for noticing and the laughs guys! These made my day! Also, ZFS is awesome! Have a good one!

Replied in thread

@var it’s nice to have a separate dedicated computer for managing the backup. You really need to have all backup disks attached to SATA. USB is not reliable for backup purposes because is likely to lose data when a sudden power outage occurs. So if you have (for example) 4 disks that all attach to SATA, that is a lot of IO happening, which you really don’t want when you are using your computer for other things that may need that IO, especially games or multimedia. You’ll notice the performance hit if it is backups are happening on a computer that you use for other things.

Also, it’s nice that the NAS comes with software that takes care of things like software RAID-1 or #ZFS for you. Some #NAS products will do #RAID1 in hardware for you.

What's the general opinion on doing scheduled zfs scrubs? Likely to be more useful on the older #zfs that #NetBSD uses?

Snapshots are taken daily and periodically cleaned up. On a system used for backing up files, does it perform a useful integrity test?

I'm messing with the ZFS array again already. When I replaced the faulted HDD earlier, it was very easy & I had no downtime, just unplug it, put a new HDD in, & do zfs replace pool [insert guid] new_device, then it automatically did everything & started online the whole time. I tried to replace a healthy HDD now, & it required taking that device offline, which makes sense, & left the array functional in a degraded state, but when unplugging it it seems like my HDDs get scrambled in /dev/, & ZFS is unable to recognize them as the disks they really are. Plugging the replacement in does not fix this, & a reboot is required before I can use zpool replace successfully. I also have to actually look up the GUID for the device to be replaced because zpool status doesn't automatically show it for the removed device like it did for the faulted device. I could do zpool status -g, but that shows the GUID for all devices instead of just the missing one, which is confusing. Maybe zpool status -gx?
#ZFS #Linux

Continued thread

Is there some magical file system/directory structure that will solve my problems? What tools should I be using? I have heard of Johnny Decimal and ZFS and more tools than I care to remember, but I don't know how to start maintaining my own data and infrastructure in a sustainable way!

johnnydecimal.com/

johnnydecimal.comA system to organise your lifeJohnny.Decimal is a system to organise your life. Find things, quickly, with more confidence, and less stress. It's free to use and the concepts are the same at home or work.
#linux#data#backup

I had an opportunity to get 200GB more on my VPS for small fee and couldn’t resist. Virtual disk has been resized but how to expand #FreeBSD #zfs zroot partition? After two minutes of digging on BSD forum I found solution - 3 commands and done! Great system and great community! (Yes I know - I like living on the edge - I have done this on the running live system 😄💥)

Replied in thread

@mwl @mms @allanjude my initial reaction was "oh g-d no way" because i have bought books that were essentially EOL'ed knowledge a year later.

but it isn't like #zfs (and hopefully not openzfs?) will have a radical rewrite or drop or change the way vdevs and zvols work or any other features so my reluctance to buy a book on technology evaporates quickly for a filesystem that solves a ton of problems like zfs has for me. i don't fully understand the poll but a book like that is worth $125 to me.

I finally got to putting together that ZFS array I wanted. It was much easier than I thought it would be. It's also extremely fast & does not appear to be having the USB slowdown problem I always have even though it's connected by USB. I'm using the five old HDDs that are "slow" & "in terrible condition" that I bought in a bundle for $25 & it's reaching almost 90MiB./s. while copying files from my big Seagate HDD, which I regularly have the USB slowdown issue with, & there's no latency when writing a ton of small files either!

Also the lights on the HDD dock blink for disk activity so it looks cool :)
#ZFS #Linux

Replied to Simon Dassow

@simondassow For ZFS, in each pool, have a non-mountable file system with a reasonable reservation. Don't actually use that to store anything; treat it as headroom for the OS and file system internals.

Because it's copy-on-write, filling up a ZFS pool completely can put you in a really tight spot.