@mhamzahkhan Hmm if "sector size" and "ashift" are interchangeable in this context, you might be borked. Which after reading the first article I linked and the bits about copying to a virtual vdev hosted on the remaining vdevs, kind of makes sense.

openzfs.github.io/openzfs-docs

@mhamzahkhan Looks like it's possible on actual Solaris ZFS.

blog.moellenkamp.org/archives/

But don't know if zpool remove is implemented in OpenZFS, I'm not at a terminal to check.

@absinthe @plants What does one do with dehydrated chillis? Is it to make a powder?

@mhamzahkhan Nope but a home enthusiast is curious to hear your conundrum 😅
I actually just finished installing Arch with ZFS root on my laptop tonight. First time doing it on a "workstation".

@mhamzahkhan Ok, well just in case it ever comes up, when someone tells you "shoot the hostage", they mean in the leg.

@mhamzahkhan In addition to RAM it will also thoroughly bludgeon your CPU. Not just on the initial data write, but also during pool scrubs. It's an enterprise feature meant for enterprise hardware to be sure. But in the case of the LTT video, that piece of kit is above and beyond even enterprise tier.

@sheogorath @mhamzahkhan I dislike any system that offers a "dumb" basic config of something, but then requires you to dive into the CLI anyway for fine tuning. (The Cockpit LibVirt module is a great example).
If you're going to end up there anyway, just start with the CLI and nothing else.
Furthermore what version of ZFS do you get in a packaged solution like unRAID/TrueNAS?
Installing the zfs-dkms module on an Arch box is one line?

@mhamzahkhan Dedup can consume a lot of ram (1-3 GB of RAM per TB of deduped data is the recommendation). But like with ARC you can set a limit on how much RAM it can consume. If the size of the dedup table exceeds this limit, dedup "spills over" to using disk. But in this case as they repeatedly stated, the disks aren't that much slower than RAM. So a spill over wouldn't cause much performance hit.

@mhamzahkhan No dedup, no special allocation class, single parity, and he didn't declare the vdev using UUID paths.
Watching ~$100k worth of equipment get configured anything less than perfectly gives me the irrates.
The guy's made his fortune, so good on him. But waving the money and gear around in the current economic environment, not to mention in front of his employees, seems a little tone deaf.

@boilingsteam if it brings more people, and by extension development, to Plasma then thumbs up 👍

@sheogorath @Dashtop This "naked scanner door", if real, would explain all the spam for male enhancement pills...

@FediFollows @nitrokey I wholeheartedly promote the use of FIDO2 wherever possible.
Unfortunately NitroKey only has one model that does FIDO2 and it only does FIDO2. So an excellent second key backup option, but not as flexible as a Yubikey ... yet. Let's hope there's more coming, we still need a FIDO2 capable key with biometrics.

@Danie Ooh good reminder to read these before the series lands.

@mhamzahkhan What's the use case? Muting notifications? I find Teams to need a lot of handholding, no real surprise really.

Michael boosted

Those who do not learn from history are doomed to repeat it: this is a must-read for the Matrix community.

RT @pierce@twitter.com

This is a great dive into how Google botched messaging for 15 years, and the table of contents BY ITSELF is a truly epic burn arstechnica.com/gadgets/2021/0

🐦🔗: twitter.com/pierce/status/1430

@PINE64 Bee tee dubs, is there still a whole crate of Pinephones hiding somewhere here in New Zealand? 🔍 🗺️ 📱

If you're using @kde on an ultrawide (32:9) monitor, then may I strongly recommend the UltrawideWindows Kwin script.
Allows you to easily snap/tile windows to the center 50% of the screen, or the left/right quarter.
I previously used a portrait-landscape-portrait orientation of three screens. This script allows me to keep my windows management style without re-training my hotkey muscle memory.

store.kde.org/p/1276605/

Show older
Mastodon

Balsillie family social hub.