ES- WARN [gc] [young]

[sunilmchaudhari (Sunil Chaudhari)]

Hi,
I am getting below warning in one of my ES node. I have 3 node cluster with 16GB heap allocated to each.
Can somebody explain the meaning of it?

[2017-05-03 11:50:40,203][WARN ][monitor.jvm ] [data-node-2] [gc][young][2507673][807406] duration [1.3s], collections [1]/[1.5s], total [1.3s]/[9.5h], memory [11.8gb]->[11.5gb]/[15.9gb], all_pools {[young] [264.8mb]->[2.7mb]/[266.2mb]}{[survivor] [24.2mb]->[33.2mb]/[33.2mb]}{[old] [11.5gb]->[11.5gb]/[15.6gb]}

What is corrective action required for it?
Cluster is in yellow state now

[polyfractal (Zachary Tong)]

It may mean nothing, or it may be indicative that your heap is getting too full. It's essentially a warning that says the young generation of heap memory took a very long time to garbage collect (1.3s).

The young gen is where new objects are created; new object creation is often transient and not long-lived, so it's stored in the young gen temporarily. Garbage collection happens frequently in the young gen, and is usually pretty fast.

If you see a few sporadic slow young gen GCs I wouldn't worry about it. You may have done a lot of heavy indexing or a big query or something to create a lot of temporary young-gen pressure.

If you're seeing this repeatedly, you may have memory pressure problems, and may want to see about either increasing total available memory or decreasing your memory footprint (evaluating queries, etc)

[sunilmchaudhari (Sunil Chaudhari)]

Hi,
Thanks for such a precise explanation.

What is meaning of below?

[1.3s], collections [1]/[1.5s], total [1.3s]/[9.5h],

memory [11.8gb]->[11.5gb]/[15.9gb]

{[young] [264.8mb]->[2.7mb]/[266.2mb]}

Can we control the space allocated for young, old and survivor space?

Any other way to invoke forceful garbage collection when young generation space runs out of memory?

br,
Sunil.

[polyfractal (Zachary Tong)]

@sunilmchaudhari

[1.3s], collections [1]/[1.5s], total [1.3s]/[9.5h],

Annotated version:

[GC took 1.3s], collections [1 GC event]/[1.5s since last GC], total [GC took 1.3s]/[9.5h spent GCing since node came online],

memory [11.8gb heap used at last GC event]->[11.5gb current heap used]/[15.9gb maximum heap used ever]

{[young] [264.8mb young gen used at last GC event]->[2.7mb young gen currently used]/[266.2mb max young gen]}