While running our app (which seemed to be running slow on this last job), I ended up receiving an rc:87 on a commit, which my app couldn't recover from. I have verified that I have free disk space, free RAM.

Upon restart, after the db was opened, the first process started and quickly reported an MDB_PAGE_NOTFOUND on a call to mdb_cursor_get

For your info, here the db info as reported during the restart: Env Stats:{ms_branch_pages=0, ms_psize=4096, ms_depth=1, ms_leaf_pages=1, ms_overflow_pages=0, ms_entries=23} Info:{me_last_pgno=678442, me_mapaddr=0, me_mapsize=4194304000, me_last_txnid=112652, me_maxreaders=300, me_numreaders=0} Data Db Stats:{ms_branch_pages=315, ms_psize=4096, ms_depth=4, ms_leaf_pages=71375, ms_overflow_pages=25, ms_entries=7768832} Primitive Db Stats:{ms_branch_pages=990, ms_psize=4096, ms_depth=4, ms_leaf_pages=183396, ms_overflow_pages=17263, ms_entries=3537190} Incidence Db Stats:{ms_branch_pages=516, ms_psize=4096, ms_depth=4, ms_leaf_pages=86349, ms_overflow_pages=0, ms_entries=11534223} % Full:66.254105%

1- What could have caused this error 87, which seems to be related to some arguments issue?

2- More importantly, what can be done to avoid getting a db in a state like this which seems to me to be corrupted (and that's also what the doc says)?

I am keeping the db should it be required for forensic analysis.

Thanks Alain