Closed stefanwouldgo closed 8 years ago
Never encountered this myself. What's your env? Test to reproduce?
Probably regression in LMDB if not observed in earlier versions.
I haven't tried reproducing, since this occured after about 12h of running bitcoin graph explorer and I wanted to get our project running ASAP again. We get random LMDB errors (like MDB_BAD_VAL) from time to time and they are never reproducible. This one is new, though.
There is an issue ITS#8336 reported on LMDB 0.9.17 (fixed in LMDB 0.9.18) which sounds similar to your problem. I would suggest you post your hardware spec, os and description of your problem to openldap-technical@openldap.org.
LMDB 0.9.18 is in lmdbjni 0.4.6 btw.
Not sure about the random MDB_BAD_VAL errors though. Might be related if the rebalance wasn't finished?
Did you have time to try 0.4.6?
We've been running 0.4.6 since then. No problems so far :)
Glad to hear. Let me know when you feel comfortable closing the issue.
You're right, this should be resolved by the fix you mentioned. So I'm closing this issue. Thank you very much for your quick and thorough help, and for making LMDB available to the JVM ecosystem!
Thanks! :-)
src/mdb.c:5276: Assertion 'NUMKEYS(mp) > 1' failed in mdb_page_search_root() Aborted
after millions of writes and reads.