FTR, this tar bug is breaking our archival process so it is ever so slightly important to me.

nomad: I have had a bit of a squiz

And I have at least some idea what's wrong

I have witnessed this squiz

I think I'm hearing good things but I'm concerned :)

nomad: I've updated the ticket

I have a test instance that I can play with without breaking production. I will be happy to test any patches that might be available. I'm in that lab every Monday & Tuesday and every other Wednesday.

nomad: If you want to try something, you can try sysmgr.org/~jclulow/tmp/tar

Let me know if it's able to archive the 5G file correctly?

gist.github.com/jclulow/ff4a2d4c50daa5df277c327529345568 is the patch, FWIW

I've started a tar c... but it takes a while to run. I'll take another look when I have some free time (right now I'm working on $job[2] and this is for $job[1])

x largefile.gz, 1074611575 bytes, 2098851 tape blocks

looks like a fail, though the extract hasn't exited yet.

oh wait, never mind. I was the one who failed.

ignore that meaningless attempt.