Here is another update. The above settings worked with labeling, but during the backups the same error popped up and broke backups. I ended up adding two more lines in SD device config to set min and max block sizes as 65536. In hindsight, I should have set it as 1 MB as per Amazon recommendation. But this works for now. Something changed with this latest tape gateway update from Amazon that enforced block sizes in virtual tapes to be multiples of 65536 (up to 1 MB) After setting those, backups have started working. I am guessing that bareos would try a range of block sizes during backup and some of them may not be multiples of default tape block sizes.
On Friday, May 3, 2024 at 6:29:46 PM UTC-3 Balagopal V . Pillai wrote: > I found the fix for the issue. Since tapeinfo confirmed 65536 to be a > valid block size, I placed Label Block Size = 65536 in the device section > of sd and that fixed the issue. It is odd that Amazon VTL worked like a > charm for two months and an update broke it. bareos-sd -f -d 100 would pump > out the current block size applied for label. Hope it helps somebody using > a similar setup. > > On Friday, May 3, 2024 at 5:42:17 PM UTC-3 Balagopal V . Pillai wrote: > >> I have a new bareos setup that uses Amazon VTL (configured with one >> drive and 10 slots to mimic a loader) that worked fine for the past 2 >> months. It looks like the tape gateway got a software update recently and >> backups have started failing today. I sent the existing virtual tapes to >> archive and started with fresh ones. Labeling gives the following error - >> >> 3307 Issuing autochanger "unload slot 2, drive 0" command. >> 3304 Issuing autochanger "load slot 1, drive 0" command. >> 3305 Autochanger "load slot 1, drive 0", status is OK. >> stored/block.cc:750 Write error at 0:0 on device "tapedrive-0" >> (/dev/nst0). ERR=Invalid argument. >> stored/block.cc:768 Write error on fd=5 at file:blk 0:0 on device >> "tapedrive-0" (/dev/nst0). ERR=Invalid argument. >> Backspace record at EOT failed. ERR=Input/output error >> 3912 Failed to label Volume: ERR=backends/generic_tape_device.cc:686 >> ioctl MTBSR error on "tapedrive-0" (/dev/nst0). ERR=Input/output error. >> >> When this happens, there is the following error in the log - >> >> [st0] Write not multiple of tape block size >> >> Here is the output of tapeinfo - >> >> Product Type: Tape Drive >> Vendor ID: 'IBM ' >> Product ID: 'ULT3580-TD5 ' >> Revision: '0103' >> Attached Changer API: No >> SerialNumber: 'CDDF10A401' >> MinBlock: 0 >> MaxBlock: 1048576 >> SCSI ID: 0 >> SCSI LUN: 0 >> Ready: yes >> BufferedMode: yes >> Medium Type: 0x48 >> Density Code: 0x0 >> BlockSize: 65536 >> DataCompEnabled: yes >> DataCompCapable: yes >> DataDeCompEnabled: yes >> CompType: 0xffffffff >> DeCompType: 0xffffffff >> BOP: yes >> Block Position: 0 >> Partition 0 Remaining Kbytes: 1042431 >> Partition 0 Size in Kbytes: 1048576 >> ActivePartition: 0 >> EarlyWarningSize: 0 >> >> >> I tried the following from bareos-sd device config and the issue >> persists - >> >> Minimum Block Size = 1048576 >> Maximum Block Size = 1048576 >> >> Tar works fine with block size of 65536 or 1048576. I was >> wondering if there is anything else I can try to fix this. Would bareos-sd >> have a debug option to print out the block size being attempted? Thanks a >> lot. >> > -- You received this message because you are subscribed to the Google Groups "bareos-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion on the web visit https://groups.google.com/d/msgid/bareos-users/a90ad0d1-cbcd-4192-8349-45dc5d1d011dn%40googlegroups.com.
