Trying to find lightweight heartbeat query to determine if a node is up in a Vertica Cluster
We have tried to use a number of different queries to serve as a heartbeat mechanism between our application and our Vertica cluster, and all seem to incur a fair amount of overhead (i.e. global catalog locking, internode communication). Have any of you found a lightweight query to a serve as a heartbeat, or, perhaps, use a different mechanism entirely to determine if a node in your Vertica cluster is up/down?
Thanks in advance
0
Comments
You can use "select 1". Or use the native connection load balancer with connection failover.
--Sharon
There are three type of heatbeats you can use,
in descending order of weight (heavy to light) and ascending order to scope (cluster, node, socket)
-- cluster check
vsql -c "select 1;"
-- node check
vsql -h node001 -c "select /* +KV(P0) */ 1;"
-- tcp check
vsql -c "\q"
Thanks a lot for the feedback.