H2O is unable to handle more than 2TB of memory

Description

Hi,

we have tried to launch H2O on an `x1e.32xlarge` EC2 instance type (4TB RAM), but ran into problem when H2O thinks it has negative memory (also breaking h2o-py in the process when it tries to print the cluster memory. Trying out -Xmx2000g seems to work, whereas anything above 2TB becomes negative.

Example `-Xmx3000g` and output of /3/Cloud endpoint - see the free_mem and max_mem keys.

Also, worth noting that we are pinned to 3.21.0.4345, because of and are anxiously waiting for a chance to upgrade.

Assignee

New H2O Bugs

Fix versions

None

Reporter

Julius Šėporaitis

Support ticket URL

None

Labels

Affected Spark version

None

Customer Request Type

None

Task progress

None

CustomerVisible

Yes

Components

Priority

Major
Configure