Checksum errors not showing affected files in raidz2 pool #17147
-
|
I have a raidz2 pool that has been experiencing checksum errors. However, when I run I have performed multiple Despite these errors, the Sometimes I can get a "errors: No known data errors" output, but still with 18 CKSUM errors. I am in zfs 2.3: And when I run How can I determine which file is causing the problem, or how can I fix the errors. Or should I just let these 18 errors exists ? |
Beta Was this translation helpful? Give feedback.
Answered by
huoxingdawang
Mar 24, 2025
Replies: 1 comment
-
|
I finally got a 0 errors, the process below may not work for everyone, but I wrote it down for the benefit of others who run into the same problem. I pasted a portion of the output of zpool events above, and I noticed that all of the zio_offsets are around 12TB (e.g., 0xc2727306000/1024/1024/1024/1024 = 12.15). Since my pool is fairly new, about a month old, and most of the operations are writing data. I assume that the data is written linearly, so zio_offsets should also be linear. I checked the logs when writing data and roughly determined which datasets were being written when writing 12T of data. I noticed that there were a few snapshots taking up a few hundred GB or so. Since I also ran some additional reads and found no additional errors were reported, I'm guessing that the errors might be happening in snapshots that are not being read. So I deleted them all. I then ran scrub and finally got 0 errors. I do remember some blogs saying their And I still don't know how those 9 errors came about, but at least |
Beta Was this translation helpful? Give feedback.
0 replies
Answer selected by
huoxingdawang
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
I finally got a 0 errors, the process below may not work for everyone, but I wrote it down for the benefit of others who run into the same problem.
I pasted a portion of the output of zpool events above, and I noticed that all of the zio_offsets are around 12TB (e.g., 0xc2727306000/1024/1024/1024/1024 = 12.15).
Since my pool is fairly new, about a month old, and most of the operations are writing data. I assume that the data is written linearly, so zio_offsets should also be linear. I checked the logs when writing data and roughly determined which datasets were being written when writing 12T of data. I noticed that there were a few snapshots taking up a few hundred GB or so. Since I also …