Thanks Kismar, that was a good post.

2018-08-28 06:49:21
 ERROR: Found chain with blocks marked invalid
2018-08-28 06:49:21 ERROR: AcceptBlockHeader: block is marked invalid
2018-08-28 06:49:21 ERROR: invalid header received 846ef0d0c305f54a821ec9061af34a4cca1319cc84654b2e2e4f777039d6a38f
2018-08-28 06:49:21 tor: Thread interrupt
2018-08-28 06:49:21 torcontrol thread exit
2018-08-28 06:49:21 Cleaning sockets...
2018-08-28 06:49:22 PrepareShutdown: Prepare Shutdown In progress...
2018-08-28 06:49:22  Shutdown - renaming thread ...
2018-08-28 06:49:22 runCommand error: system(./biblepayd -erasechain) returned 32512


So it is revealing.  On my end I need to take a look at something - it appears we do sort of jump the gun in the "Found chain with blocks marked invalid".  We supposedly attempt to replay the last 205 blocks when that happens, but on your node it appears we jumped down and tried to restart it with -erasechain (see last line in the log) and that command returned an error (error 32512).

So on your end, if you are starting biblepay from an absolute path, (like from a script) that would explain 32512.  Ill take a look at this.  If you want autorecover to work, you could modify your script to :  unalias cd && cd ~/biblepay/src/qt && ./biblepayd     when it starts and then it should recover when this happens.  But this is just for the time being as we make this fullproof and resilient.