Opened 4 years ago
Last modified 5 weeks ago
#16830 new enhancement
BFS Needs to be more Robust on Corrupted Disks
| Reported by: | AGMS | Owned by: | axeld |
|---|---|---|---|
| Priority: | normal | Milestone: | Unscheduled |
| Component: | File Systems/BFS | Version: | R1/beta2 |
| Keywords: | Cc: | agmsmith@… | |
| Blocked By: | Blocking: | ||
| Platform: | All |
Description
Often when encountering corrupted data on disk, BFS will do a kernel panic on some virtual memory page in the cache. It would be nicer if it could recover from that sort of error or test for page validity before the VM tries to access it.
Many times checkfs doesn't fix it and thus a reformat is the most reliable way to make it go away.
Attachments (1)
Change History (2)
by , 4 years ago
| Attachment: | CorruptBFSWriteVMPanic20210305.jpeg added |
|---|
comment:1 by , 5 weeks ago
| Keywords: | robust removed |
|---|
Any reliable way of reproducing this? I can't recall seeing it.
Note:
See TracTickets
for help on using tickets.
A kernel panic on a corrupted disk while an innocent program tries to write to a log file.