11-12-2016 10:14 AM - last edited on 11-13-2016 07:11 AM by cjervis
I added Kudu service to my cluster, When tried to start the service i got the following error:
+ exec /opt/cloudera/parcels/KUDU-1.0.1-1.kudu1.0.1.p0.14
/lib/kudu/sbin/kudu-master --flagfile=/var/run/cloudera-scm-agent/process/301 2-kudu-KUDU_MASTER/gflagfile F1112 19:48:48.629565 1939 master_main.cc:59] Check failed: _s.ok() Bad status: Service unavailable: Cannot initialize clock: Error reading clock. Clock considered unsynchronized
Checked the ntpd service and all is fine and the clock is synced at all nodes.
11-14-2016 08:24 PM
11-14-2016 08:54 PM
I already followed this blog, ntpd service is fine and the clock is synced at all nodes.
Yes, it's on all the nodes including the master, seemd Kudu is more senstive to the clock sync issue from other services, Where i can config for example the --max_clock_sync_error_usec
11-15-2016 10:18 AM
You can configure --max_clock_sync_error_usec in /etc/kudu/master.gflagfile or /etc/kudu/tserver.gflagfile. There should be existing files in there with a few configurations; you can add more.
I hope that helps. The error message you're seeing doesn't seem related to the max error (if it were, it'd include something like "error: 11130000, is past the maximum allowable error: 10000000".