Elastic framework and it's tasks memory defaults are problematic

[bergerx]

Can you parameterize the memosry for elastic framework?

It's currently set to 2G (https://github.com/mesosphere/universe/blob/version-3.x/repo/packages/E/elastic/5/marathon.json.mustache#L4) but in out 1.9 cluster the elastic marathon task keeps restarting by OOM. I bumped the mem in marathon task defition to 3G to solve the problem.

There is also similar problem with the proxylite task started by elastic framework, but i can tweak it by updating PROXYLITE_MEM env value from the default 100 to 200. Even though restarts, proxylite does its job mostly, but I expect everybody using the framework are experiencing this, tweaking the default could be better.

[bergerx]

others are also reporting: https://dcos-community.slack.com/archives/C132NTZJN/p1492704511646400

Ingest's defaults are also does not seem good.here is another one hit issue with that: https://dcos-community.slack.com/archives/C132NTZJN/p1492780815802640 Here is a screenshot to see increasing memory for ingest did help:

[bergerx]

Changes I did:

• Increased mem for Framework task on marathon from 2G to 3G
• Increased memory of elastic framework using environment values of the framework:
  • PROXYLITE_MEM : 100 --> 200
  • INGEST_NODE_MEM: 1G --> 2G (this change is likely overkill)