Forum Stats

  • 3,769,496 Users
  • 2,252,970 Discussions
  • 7,875,056 Comments

Discussions

Notice: Workaround needed for out-of-disk problem in BDB JE 7.0 and earlier

This notice applies to JE HA applications only.

To avoid a potential out-of-disk problem when one node in a replication group is down for more than 30 minutes, make the following configuration change. Without this change, other nodes in the group will retain data files for up to 24 hours. If data files are retained for a long time period, and the application performs writes at a high rate, running out of disk on the other nodes in the group is likely.

Via the API, make the following call when configuring your application:
ReplicationConfig.setConfigParam(ReplicationConfig.REP_STREAM_TIMEOUT, "30 min");
Via the je.properties file, add the line:
je.rep.repStreamTimeout = 30 min

Alternately, applications may upgrade to JE 7.1 or higher. In JE 7.1 the default value for REP_STREAM_TIMEOUT was changed from 24 hours to 30 minutes. In JE 6.0, the default value in the documentation for REP_STREAM_TIMEOUT was changed to 30 minutes, but the default value in the product was not actually changed until JE 7.1.

-- The JE team

This discussion has been closed.