spacecowboy @spacecowboy17.bsky.social · 21/02/2026For You will be down for ~1-2 hours as I replace the system drive. The Kingston NV2 drive has been freezing up the computer - often for a minute at a time and sometimes indefinitely until a manual restart. I'm replacing it with a Samsung 990 Pro, which should be more stable. 1422905161
spacecowboy @spacecowboy17.bsky.social · 21/02/2026Hm, running "dd if=/dev/nvme2n1 of=/dev/nvme1n1 bs=64M conv=fsync status=progress" and it is super slow 40MB/s Copied 60GB out of 2TB. "iostat -x 1" shows that the source drive is 100% utilized with no activity. Suggestions welcome 111322
JauntyWunderKind @jauntywunderkind.bsky.social · 21/02/2026Hopefully not here yet but, conv=sync,no error maybe might possibly let only a little corruption happen, if things get dire. Are you watching smart or kernel logs at all? superuser.com/questions/62...superuser.comWhat does dd conv=sync,noerror do?What's the effect of adding conv=sync,noerror when backing up an entire hard disk onto an image file? Is conv=sync,noerror a requirement when doing forensic stuff? If so, why is it the case with 130
nikomo @nikomo.fi · 21/02/2026Wonder if it's the fsync in combination with the massive blocksize. I'd get other opinions but personally I'd drop both options and just let dd do its thing, you can sudo sync afterwards to flush caches. 140
Russ The Tedious Ghost @russ41.bsky.social · 21/02/2026Kinda wondering if the computer freezing and the slow transfer don't share a root cause... 050
Kernel @foxden.tech · 22/02/2026Pray to the Omnissiah, cantations and offering to the Machine God will guide you brother. 000
Marc Slemko @znep.com · 21/02/2026100% utilized is a trick because that shows at least one single queued IO operation, but you need to keep the request queues full on a SSD for parallelization behind the scenes, so you want to look at those numbers. Posting the output of "iostat -xdm 1" over a minute for both drives may give clues. 1110
Maddie @raw.rs · 22/02/2026partclone can be way faster than DD if there’s a lot of empty space as it reads the file table and only copies what’s actually in use. 030
Smingers Diddit @smingers.bsky.social · 21/02/2026conv=fsync could be what's slowing it down. unless you need it for a specific reason you could try omitting that arg ddrescue might be worth looking into if there might be a dying drive involved (you can resume it, saves a log etc) 100
spacecowboy @spacecowboy17.bsky.social · 22/02/2026It looks like I can't make it much faster than 200MB/s Ended up running "sudo cat /dev/nvme2n1 | sudo tee /dev/nvme1n1 > /dev/null" - it seems to utilize the reading part quite well with prefetch of the filesystem cache. Will let it finish. May take 2-3 hours though 11140