New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[FLINK-934] Remove default values for JVM heap sizes in startup scripts #14
Conversation
…heap sizes in startup scripts
Robert Metzger on dev@flink.incubator.apache.org replies: |
Okay, that looks good. The other flags (max perm gen size, mark sweep GC, mark sweep class unloading) are unaffected by that? |
When we pass the flags, we make sure that Xms and -Xmx are the same, to prevent the JVM from being funny and resizing the heap from time to time. I think we loose that with this option. But I guess its all right, on cluster setups we would expect those to be set anyways. |
Yes, the other flags are set as before. The While doing the PR, I came to the same conclusion as your second comment. But we can actually also do something more fancy and figure out the available memory ourselves and set it (thereby circumventing the defaults). |
Let's not do this ourselves. I think it is fine. If you want a robust setup, set the values. Our YARN scripts also set them automatically, based on the container size, afaik. |
I will merge this... |
# This is the 1st commit message: FLINK-8255 Key expressions on named row types do not work Test case from description + 2 more Resolution of class cast by using of TupleTypeInfoBase rather than TupleTypeInfo # This is the commit message #2: FLINK-8255 Key expressions on named row types do not work Test case from description + 2 more Resolution of class cast by using of TupleTypeInfoBase rather than TupleTypeInfo # This is the commit message #3: FLINK-8255 Key expressions on named row types do not work Test case from description + 2 more Resolution of class cast by using of TupleTypeInfoBase rather than TupleTypeInfo # This is the commit message #4: FLINK-8255 Key expressions on named row types do not work Test case from description + 2 more Resolution of class cast by using of TupleTypeInfoBase rather than TupleTypeInfo # This is the commit message #5: FLINK-8255 Key expressions on named row types do not work Test case from description + 2 more Resolution of class cast by using of TupleTypeInfoBase rather than TupleTypeInfo # This is the commit message #6: FLINK-8255_Key_expressions_on_named_row_types_do_not_work corrections based on comment https://issues.apache.org/jira/browse/FLINK-8255?focusedCommentId=16466381&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-16466381 # This is the commit message #7: FLINK-8255_Key_expressions_on_named_row_types_do_not_work added expected exception type # This is the commit message #8: FLINK-8255_Key_expressions_on_named_row_types_do_not_work omit issue not related changes # This is the commit message #9: FLINK-8255_Key_expressions_on_named_row_types_do_not_work use instanceof as org.apache.flink.api.java.typeutils.RowTypeInfo#isTupleType leads to org.apache.flink.table.api.batch.ExplainTest#testJoinWithoutExtended failure # This is the commit message #10: Revert "FLINK-8255_Key_expressions_on_named_row_types_do_not_work" This reverts commit 497aa20. # This is the commit message #11: Revert "FLINK-8255_Key_expressions_on_named_row_types_do_not_work" This reverts commit d891cc5. # This is the commit message #12: FLINK-8255_Key_expressions_on_named_row_types_do_not_work omit issue not relate changes # This is the commit message #13: FLINK-8255_Key_expressions_on_named_row_types_do_not_work added expected exception to tests # This is the commit message #14: FLINK-8255_Key_expressions_on_named_row_types_do_not_work added instanceof checks for DataSet # This is the commit message #15: FLINK-8255_Key_expressions_on_named_row_types_do_not_work Tests corrected
…ease scripts This closes apache#14.
* Introduce manifest readers and writers * Fix dependency convergence test * Fix comments * Skip dependency convergence test
add missing licenses
* Add Flink client configurations * Add copy command to copy the required dependencies in lib/ * adjust change-version.sh * Update sql-client-defaults.yaml
…ta (apache#14) DE-1283: Pipeline report non-zero bytes out when there is no input data
This is a fix for FLINK-934. As a follow up we should have the heap config keys only as suggested entries (commented out) in FLINK-898 aka #6.
With
jobmanager.heap.mb
andtaskmanager.heap.mb
not set, 16 GB of main memory and Java 7.JobManager with 4294967296/2^20 = 4096 MB heap size.
$ bin/start-cluster.sh Starting job manager Starting task manager on host batman $ jps 10049 TaskManager 9966 JobManager $ jinfo -flag MaxHeapSize 9966 -XX:MaxHeapSize=4294967296 $jinfo -flag MaxHeapSize 10049 -XX:MaxHeapSize=4294967296
Both JobManager and TaskManager with 4294967296/2^20 = 4096 MB heap size each.
Setting
jobmanager.heap.mb
andtaskmanager.heap.mb
to 256 and 512 MB respectively.JobManager with 805306368/2^20 = 768 MB (256 for the JobManager and 512 MB for the TaskManager).
$ bin/start-cluster.sh Starting job manager Starting task manager on host batman $ jps 10833 JobManager 10918 TaskManager $ jinfo -flag MaxHeapSize 10833 -XX:MaxHeapSize=268435456 $jinfo -flag MaxHeapSize 10918 -XX:MaxHeapSize=536870912
JobManager with 268435456/2^20 = 256 MB and TaskManager with 536870912/2^20 = 512 MB.