Skip to content
This repository has been archived by the owner on Jan 13, 2022. It is now read-only.

Commit

Permalink
add seqNo between Corona cluster manager and CoronaJobTracker
Browse files Browse the repository at this point in the history
Summary:
seqNos are added to cluster manager and corona jobtracker transactions. heatbeat from JT will not change the seqNo. CM will
save the seqNo from JT, and expect that will be the next value from JT for any transaction other than heartbeta. if the seqNo
doens't match, CM will kill the JT. JT will save the seqNo from CM, and expect that will be the next value from CM. If the seqNo
doesn't match, JT will kill itself.

Test Plan:
the sleep sample(safeMode and normal mode), TestSessionDriver, TestLoadManager, TestClusterManager, TestPreemption,
TestResourceCheck

Reviewers: rvadali, aching

Reviewed By: rvadali

Task ID: 1794271
  • Loading branch information
jeanxu authored and Alex Feinberg committed Nov 8, 2012
1 parent 1cee40e commit 238c612
Show file tree
Hide file tree
Showing 13 changed files with 1,662 additions and 197 deletions.
5 changes: 5 additions & 0 deletions singleNodeHadoop/coronaConf/mapred-site.xml
Original file line number Diff line number Diff line change
Expand Up @@ -5,6 +5,11 @@

<configuration>

<property>
<name>cm.heartbeat.delay.max</name>
<value>300000</value>
</property>

<property>
<name>cm.config.reload.period.ms</name>
<value>5000</value>
Expand Down
7 changes: 7 additions & 0 deletions src/contrib/corona/interface/ClusterManager.thrift
Original file line number Diff line number Diff line change
Expand Up @@ -127,6 +127,11 @@ struct SessionInfo {
9: optional i64 deadline
}

struct HeartbeatArgs{
1: required ResourceRequestId requestId,
2: required ResourceRequestId grantId,
}

struct ClusterManagerInfo {
1: required string url,
2: required string jobHistoryLocation,
Expand Down Expand Up @@ -193,6 +198,8 @@ service ClusterManagerService {
// Heartbeat a session.
void sessionHeartbeat(1: SessionHandle handle) throws (1: InvalidSessionHandle e, 2: SafeModeException f),

void sessionHeartbeatV2(1: SessionHandle handle, 2: HeartbeatArgs heartbeatArgs) throws (1: InvalidSessionHandle e, 2: SafeModeException f),

// Request additional resources. A request is required for each resource
// requested.
void requestResource(1: SessionHandle handle, 2: list<ResourceRequest> requestList) throws (1: InvalidSessionHandle e, 2: SafeModeException f),
Expand Down
Loading

0 comments on commit 238c612

Please sign in to comment.