Reputation
12,442
Top tag
Next privilege 15,000 Rep.
Protect questions
Badges
1 16 28
Newest
 hard-disk
Impact
~411k people reached

18h
comment Encrypt existing disk with LUKS?
Unless it's LVM all data has to be moved to make room at the beginning for the LUKS header (2MiB). So it's not a straightforward / risk-free process. Tell us more about your setup (Partitions/RAID/LVM/...)
Jul
27
comment SSD failiure. grub rescue sees files, but I can't mount drives
If the device itself is bad, you should probably start with ddrescue.
Jul
27
comment Is it possible, with `sed` or otherwise, to apply a regex transform to an input stream that may not contain newlines?
Is the source plain ascii or random binary data? Does the source protocol tell you where those s:x:y:g are going to show up? I recently had this issue with multipart/form-data. I made it work with sed but it was so ugly and slow that in the end, I used a parser written in C.
Jul
27
comment Does Gparted change permissions?
A newly mkfs'd Linux filesystem is owned by root:root by default.
Jul
25
comment Is there any way to prevent a bad drive from disappearing from /dev?
lspci, and if it's AHCI, ls /sys/bus/pci/drivers/ahci/ would have links to individual controllers and they contain links to ataX...
Jul
25
answered Is there any way to prevent a bad drive from disappearing from /dev?
Jul
24
comment SSD: how often should I do fstrim?
Google ssd endurance test, there are several. Most SSD can handle more writes than you will ever need (outside of heavy duty database server or something). Of course it may still fail for various reasons, there is no reliable storage, you have to make backups in any case.
Jul
24
comment SSD: how often should I do fstrim?
It's related, of course, but strictly bound? TRIM helps performance even in cases that do not involve write amplification. If you overwrite a full erase block worth of data there is no write amplification (you have to write the whole thing either way). But if there are no free blocks the SSD has to erase one first and erasing is considerably slower than writing an already erased block.
Jul
24
revised SSD: how often should I do fstrim?
added 49 characters in body
Jul
24
answered SSD: how often should I do fstrim?
Jul
22
comment /dev/mmcblk0 partitions limit
But adding a kernel parameter seems the simplest thing all things considered.
Jul
22
comment /dev/mmcblk0 partitions limit
Other devices have partition limits as well (usually 15 instead of 7). There is 259 block Block Extended Major (not sure why not used by mmc, probably no one expected it) but even that has limits. You can use partx, kpartx, and map additional partitions through device mapper w/o using "LVM" per se. LVM just makes things simpler, if you're going to have to use the device mapper anyway. In an embedded environment that does not have device mapper, losetup with offset/sizelimit might also be an option.
Jul
22
revised /dev/mmcblk0 partitions limit
added 123 characters in body
Jul
22
answered /dev/mmcblk0 partitions limit
Jul
15
answered What is the difference between a directory name that ends with a slash and one that does not?
Jul
15
reviewed Approve Problems with creating lotto ticket
Jul
14
revised What's a smart way to count the number of days since X?
added 35 characters in body
Jul
14
answered What's a smart way to count the number of days since X?
Jul
13
comment How to replace faulty device on RAID5
This is not something that happens by cloning a disk. It could have been a --grow --raid-devices=7. In which case you still have a 7 disk raid, as reported by the metadata of all disks. Look at the array size: 18000GB / 3000GB = 6 means a 7 disk RAID5 (6 data + 1 parity). Before doing anything else you should verify (in read-only mode) that your data is still there.
Jul
13
comment How to replace faulty device on RAID5
Your RAID firmly believes that it's supposed to have 7 devices. So there is in reality one drive missing. Even so it should run & give you data access with 6 out of 7 devices.