Forum: Installation

Post: ntp problem when starting voltdb as cluster

ntp problem when starting voltdb as cluster
Sabrina
Mar 30, 2014
Hello,

I have two ubuntu servers (192.168.10.21 and 192.168.10.22).
On both servers I do this:
/etc/init.d/ntp stop
ntpdate europe.pool.ntp.org
/etc/init.d/ntp start

The time is on both systems correct. When I start the cluster with this command:
/opt/voltdb/bin/voltdb create helloworld/tpcc.jar --deployment=deployment.xml --host=192.168.10.21

- on the first server (192.168.10.21) I get this output:
Build: 4.0.2.3 voltdb-4.0.2.3-0-ge2f2df4-local Enterprise Edition
Connecting to VoltDB cluster as the leader...
Host id of this node is: 0

- on the second server (192.168.10.22) I get this output:
Build: 4.0.2.3 voltdb-4.0.2.3-0-ge2f2df4-local Enterprise Edition
Connecting to the VoltDB cluster leader /192.168.10.21:3021
FATAL: Clock skew is 129 which is > than the 100 millisecond limit. Make sure NTP is running.
VoltDB has encountered an unrecoverable error and is exiting.
The log may contain additional information.

Can anyone help me, why the time is inccorrect? I have no ideas. Please help me.

Thanks.

Best regards,
Sabrina
nshi
Mar 31, 2014
Hi Sabrina,

Syncing with an external NTP source provides you with relatively accurate time for most applications, but machines may still have some small drifts between each other. In this case, the drift was 129 milliseconds.

To reduce the drift between servers, you can set up one of the servers as the source and point the rest of the servers to it. The following link shows you how to do this with an example configuration at the bottom of the page.

https://voltdb.com/docs/PerfGuide/ntpSvcIntro.php