AnsweredAssumed Answered

Hung threads

Question asked by son9673 on Dec 28, 2015
Latest reply on Jan 20, 2016 by son9673
We're restoring MapR DB backups to a new table, and everything goes well but periodically the map-reduce jobs will hang forever and ultimately fail when tasks time out (after several hours). We've noticed that we can detect when the job is hung by running

    /opt/mapr/server/mrconfig dbinfo threads

And if there are any waiters then it's a sign that something is hung. We've been restarting warden on the node that has waiters, and that's made our hung jobs complete successfully.

Why do threads get hung? How can we avoid this problem?