Conversation
Notices
-
wow - badblocks test with read-write testing (4 patterns) across the whole volume has finished. result: "Pass completed, 1 bad blocks found. (1/0/0 errors)". Seemingly all that writing (also scrub before this test) has done some good. \o/ I fully expected that NOT to happen... of course, they might re-appear... and I still see IO errors in dmesg - huh??
-
@mk so: new strategy: split the disk into two volumes: one including the (former) area of bad blocks at the end, and one starting after that. If teh bad blocks re-appear in the *same* area (more or less) I can resize (shrink) the volume to avoid them. And, of course, use the disk only for fast access to 'scratch' data that always gets backed up anyway.
-
@mk that should be 'across the whole disk' of course (it's scrubbed, there *is* no volume on there now)
-
@mk looking at the old reports (ddrescue) I could even split that into three volumes, the first two *ending* with a bad of iffy area
-
-
@mk not so easy: bad blocks are 4096b blocks on the raw disk; creating volumes I get the maximum size in MiB which alrady takes account of the overhead for the FS (and partition table) itself; ended up estimating based on percentages #ugh. bad memories of doing such calculations before, long ago... #ihatehardware
-
-
-
