[ https://issues.apache.org/jira/browse/TWILL-216?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15872256#comment-15872256 ]
ASF GitHub Bot commented on TWILL-216: -------------------------------------- Github user yufeldman commented on a diff in the pull request: https://github.com/apache/twill/pull/33#discussion_r101817212 --- Diff: twill-core/src/main/java/org/apache/twill/internal/json/TwillRuntimeSpecificationCodec.java --- @@ -94,6 +96,8 @@ public TwillRuntimeSpecification deserialize(JsonElement json, Type typeOfT, jsonObj.has(RM_SCHEDULER_ADDR) ? jsonObj.get(RM_SCHEDULER_ADDR).getAsString() : null, logLevels, - maxRetries); + maxRetries, + jsonObj.has(HEAP_RESERVED_MIN_RATIO) ? --- End diff -- sure > Make ratio between total memory and on-heap memory configurable > --------------------------------------------------------------- > > Key: TWILL-216 > URL: https://issues.apache.org/jira/browse/TWILL-216 > Project: Apache Twill > Issue Type: Improvement > Components: yarn > Reporter: Yuliya Feldman > Assignee: Yuliya Feldman > > As of now ratio between on-heap memory and total memory provided to yarn > container is hardcoded to 0.7, so if app running in the container needs more > reserved memory than on-heap it is not possible to achieve. > Suggestion is to make it configurable as well as amount of reserved memory -- This message was sent by Atlassian JIRA (v6.3.15#6346)