ES 2.1.1 translog grows forever

We are performing many writes and updates in our ES 2.1.1 cluster, and within ~2 weeks time the translog reaches 50GB on disk, and continues to grow seemingly indefinitely.

Is this a known issue? was it fixed in newer versions? can this be fixed on 2.1.1 without upgrading to a newer version for now?


Yes, this is a known bug fixed in 2.1.2 and 2.2.0. You can issue flush commands to clear and thus workaround until you can upgrade.