Am 30.11.20 um 00:40 schrieb Nathan Stratton Treadway:
On Thu, Nov 26, 2020 at 11:12:45 +0100, Stefan G. Weichinger wrote:
I was able to amrestore correctly .. no "-r", no skipping.
(Did you try this while the real pigz binary was in place, or only after
you replaced it with "gzip"?)
I think with the real pigz, but not 100% sure right now.
Since you have a workaround now I don't know how much more effort you
want to spend on this, but if you do want to investigate further you
could try replacing /usr/bin/pigz with a shell script wrapper which
calls pigz-original but writes some debugging messages, etc. to a log
file before and after invoking pigz-original, and exits with a "success"
status. Basically just trying to see if there is any fussing you can do
to how pigz is invoked or how the exit status is processed which changes
the overall behavior of amrestore-calling-pigz.
This depends on my overall workload.
The customer won't pay me for debugging that, I assume.
I also thought of a wrapper forking to gzip when called at decompression
time.
And I filed an issue:
https://github.com/madler/pigz/issues/80
