Archives are important. When they are public and available for searching, it retains and passes on knowledge. It saves vast amounts of time.
Case in point
I started the copy-backups-to-tape process today.
This appeared on the tape server:
Jan 7 19:12:08 r720-01 kernel: (sa0:mps0:0:5:0): 64512-byte tape record bigger than supplied buffer
Damn. Do I have a tape problem?
Search results
When searching for this, I found this FreeBSD Forums post from 2016 where I learned it is not a problem. It is a test; it is information, not an indication of an error.
But wait, there’s more
What was fun for me, that thread refers to a FreeBSD SCSI mailing list post from 2015 where Ken Merry states:
Yes, it’s informational. It tells you that your tape blocks are 64512
bytes long. Or at least the first one is.The initial tape mount inside the sa(4) driver does a test read with an 8K
buffer. This is to get the drive to actually look at the media, so it will
know what is there. (This is necessary on some older drives.)
The irony
The irony is that Ken was answering my question on this same topic. I had completely forgotten about this issue.
Blog, post, record it
Please blog if you can.
If you post to a mailing list and then later find the solution, reply to your original email with the answer.
If you post to a forum, do the same if you find the answer later.
Future you will thank you.
