Re: Seemingly erratic behavior

[Rick James]

Let's focus for a minute on the Xtradb Cluster...

You say there are 2 "nodes"? (3 could give you HA; 2 is hardly worth having.)

The nodes are in VMs? On the same machine? On different machines in the same datacenter? In geographically separate locations?

The last case (geo dispersed) is good for HA, but poor for performance. I will assume this is the root cause of the timing difference...

Let's talk about transactions. Do you code with BEGIN (or START) and COMMIT? Or do you use autocommit? What value for autocommit?

But... That has to do with writes. Your query is a SELECT? Could it be inside a transaction?

Cross-country transactions will cost extra time at COMMIT. (We are talking tens of millisecond.)

Well, it seems unlikely that Xtradb Cluster is the 'problem', but it is good to keep it in mind.