Are there any exception stack traces in the log when the processors fail / before that?
On Thu, Oct 22, 2020 at 12:28 AM <jw4306...@gmail.com> wrote: > Hello! > > > > We’re running into a problem with NiFi 1.11.4. > > > > Our HBASE/HDFS/Parquet processors are all configured with a master > KeytabCredentialsService that is pointing to a Kerberos principal and > keytab file. > > > > The environment’s /etc/krb5.conf file has the line renew_lifetime = 7d > commented out due to an issue with Java-OpenJDK (that is apparently fixed > but still shows up) causing “MESSAGE STREAM MODIFIED (41)” errors to appear > whenever we have it uncommented. > > > > When NiFi starts, it is able to kinit with the Kerberos KDC and is issued > a 24 hour ticket. Everything works fine right up until that ticket expires. > Once the ticket expires, all of our HDFS/HBASE/Parquet processors start > failing. > > > > I haven’t been able to find anything in our logs around the timeframe, but > I can’t turn on debug logging for this because the logs are tremendously > large when we do that (approximately 100-200 MB per minute and the problem > only occurs at the 24 hour mark). > > > > How would we go about troubleshooting this issue? > > > > Environment: > > Red Hat Enterprise Linux 7.9 > > Apache NiFi 1.11.4 > > java-11-openjdk 11.0.8.10-1.el7 > > > > Thanks! > > > > >
