That message has existed in the code for at least 15 years, and I think
this is the first time anyone has reported seeing it. Likely, the thread
crashed with an exception, but unfortunately it currently doesn't log
the exception message.
The LZMA SDK was upgraded in 6.4.0, and there were updates to the code
on the IS side that interfaces with the SDK. Perhaps a new bug was
introduced somewhere.
Which version of IS are you running? If >=6.4.0, did the error only
start after upgrading to 6.4?
You mentioned that you were using "lzma2/max". Is Compression=lzma2/max
the *only* compression-related setting you are using? i.e. no LZMA*
settings?
You mention that it works "sometimes". Does the error seem to appear
randomly? If you do 10 builds in a row, do some succeed and some fail,
and when it does fail, does it fail in a different place each time? It's
too early to say for sure, but if it is totally random, then that
*could* indicate a hardware problem like bad RAM. (Have other apps been
crashing mysteriously?)
Can you try setting LZMAUseSeparateProcess=yes? Does the error message
change?
6.4.3 will log the exception message (if any) which ought to help
diagnose what's causing the thread to terminate.
-JR