Unexplained FD->SD disconnects (block size?)

[Ruth Ivimey-Cook]

Folks,

I am seeing unexplained disconnects between FD and SD for some (a minority) of my backups, though perhaps more correlated with host than job. FD is on Debian while SD is Ubuntu, both amd64. This particular error was several hundred gigs into a job and >30G into the current LTO5 tape.

The logs look like this:

17-Aug 00:59 helva-sd JobId 17163: Moving to end of data on volume "Incr-105"
17-Aug 00:59 helva-sd JobId 17163: Ready to append to end of Volume "Incr-105" at file=1.
17-Aug 00:59 helva-sd JobId 17163: New volume "Incr-105" mounted on device "LTODrive4" (/dev/tape/by-id/scsi-HU1226P1P0-nst) at 17-Aug-2025 00:59.
17-Aug 01:03 helva-sd JobId 17163: Despooling elapsed time = 00:06:21, Transfer rate = 135.2 M Bytes/second
17-Aug 01:03 helva-sd JobId 17163: Spooling data again ...
17-Aug 01:14 helva-sd JobId 17163: User specified Job spool size reached: JobSpoolSize=51,539,738,579 MaxJobSpoolSize=51,539,607,552
17-Aug 01:14 helva-sd JobId 17163: Writing spooled data to Volume. Despooling 51,539,738,579 bytes ...
17-Aug 01:14 helva-sd JobId 17163: Fatal error: stored/spool.cc:456 Spool block too big. Max 64512 bytes, got 524288
17-Aug 01:14 helva-sd JobId 17163: Despooling elapsed time = 00:00:01, Transfer rate = 51.53 G Bytes/second
17-Aug 01:14 helva-sd JobId 17163: Releasing device "LTODrive4" (/dev/tape/by-id/scsi-HU1226P1P0-nst).
17-Aug 01:14 helva-sd JobId 17163: Elapsed time=04:41:03, Transfer rate=56.90 M Bytes/second
17-Aug 01:14 helva-fd JobId 17163: Error: lib/bsock_tcp.cc:469 Wrote 236589 (mlen: 236585) bytes to Storage daemon:helva.cam.ivimey.org:9103, but only 65536 accepted.
17-Aug 01:14 helva-fd JobId 17163: Fatal error: filed/backup.cc:1413 Network send error to SD. ERR=Connection reset by peer
17-Aug 01:14 helva-fd JobId 17163: Warning: Encountered 1 xattr errors while doing backup
17-Aug 01:14 greyarea-bareos-dir JobId 17163: Error: Bareos greyarea-bareos-dir 24.0.0~pre1546.c16dbcf30 (14Dec24):

Does anyone have any idea what is going on?

• FD: 24.0.1~pre67.4ee24a825 (12 February 2025),
• SD/DIR: 24.0.0~pre1546.c16dbcf30.

Would updating both to current=24.0.4 help?

Ruth

[Bruno Friedmann (bruno-at-bareos)]

Hi Ruth,

Would it be possible to get the whole joblog, and also, could you report if any other jobs were using the same device, and what is the definition of that tape device ( LTODrive4 )

Check into logs, if at the same time, on this device or volume an error occur, or whatever incident you can find.

for us from what we see, it might be a device has it block size set to 63k for a label, and then failed to reset the block size. 512k is the normal size for spool ...