Skip to content

ZOOKEEPER-4257: learner.asyncSending and learner.closeSocketAsync should be configurable in zoo.cfg #1645

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.

Sign up for GitHub

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

Closed
ayushmantri wants to merge 2 commits into from
Closed

ZOOKEEPER-4257: learner.asyncSending and learner.closeSocketAsync should be configurable in zoo.cfg #1645

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
5929312
ZOOKEEPER-4257: learner.asyncSending and learner.closeSocketAsync sho…
ayushmantri Mar 17, 2021
8ff5523
ZOOKEEPER-4257: learner.asyncSending and learner.closeSocketAsync sho…
ayushmantri Mar 25, 2021
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
20 changes: 14 additions & 6 deletions zookeeper-docs/src/main/resources/markdown/zookeeperAdmin.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1151,14 +1151,22 @@ property, when available, is noted below.
When set to 0, no requests will be throttled. The default is 0.

* *learner.closeSocketAsync*
(Jave system property only: **learner.closeSocketAsync**)
**New in 3.6.2:**
When enabled, a learner will close the quorum socket asynchronously. This is useful for TLS connections where closing a socket might take a long time, block the shutdown process, potentially delay a new leader election, and leave the quorum unavailabe. Closing the socket asynchronously avoids blocking the shutdown process despite the long socket closing time and a new leader election can be started while the socket being closed. The default is false.
(Jave system property: **zookeeper.learner.closeSocketAsync**)
**New in 3.7.0:**
When enabled, a learner will close the quorum socket asynchronously. This is useful for TLS connections where closing a socket might take a long time, block the shutdown process, potentially delay a new leader election, and leave the quorum unavailabe. Closing the socket asynchronously avoids blocking the shutdown process despite the long socket closing time and a new leader election can be started while the socket being closed.
The default is false.

* *leader.closeSocketAsync*
(Java system property only: **leader.closeSocketAsync**)
**New in 3.6.2:**
When enabled, the leader will close a quorum socket asynchoronously. This is useful for TLS connections where closing a socket might take a long time. If disconnecting a follower is initiated in ping() because of a failed SyncLimitCheck then the long socket closing time will block the sending of pings to other followers. Without receiving pings, the other followers will not send session information to the leader, which causes sessions to expire. Setting this flag to true ensures that pings will be sent regularly. The default is false.
(Java system property: **zookeeper.leader.closeSocketAsync**)
**New in 3.7.0:**
When enabled, the leader will close a quorum socket asynchoronously. This is useful for TLS connections where closing a socket might take a long time. If disconnecting a follower is initiated in ping() because of a failed SyncLimitCheck then the long socket closing time will block the sending of pings to other followers. Without receiving pings, the other followers will not send session information to the leader, which causes sessions to expire. Setting this flag to true ensures that pings will be sent regularly.
The default is false.

* *learner.asyncSending*
(Java system property: **zookeeper.learner.asyncSending**)
**New in 3.7.0:**
When enabled, sending and receiving packets in learners are done asynchronously. Slow network between leader and follower hangs the follower, to avoid this problem enable this feature.
The default is false.

* *forward_learner_requests_to_commit_processor_disabled*
(Jave system property: **zookeeper.forward_learner_requests_to_commit_processor_disabled**)
Expand Down
4 changes: 2 additions & 2 deletions zookeeper-server/src/main/java/org/apache/zookeeper/server/quorum/Learner.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -118,9 +118,9 @@ public Socket getSocket() {

private static final boolean nodelay = System.getProperty("follower.nodelay", "true").equals("true");

public static final String LEARNER_ASYNC_SENDING = "learner.asyncSending";
public static final String LEARNER_ASYNC_SENDING = "zookeeper.learner.asyncSending";
private static boolean asyncSending = Boolean.getBoolean(LEARNER_ASYNC_SENDING);
public static final String LEARNER_CLOSE_SOCKET_ASYNC = "learner.closeSocketAsync";
public static final String LEARNER_CLOSE_SOCKET_ASYNC = "zookeeper.learner.closeSocketAsync";
public static final boolean closeSocketAsync = Boolean.getBoolean(LEARNER_CLOSE_SOCKET_ASYNC);

static {
Expand Down
2 changes: 1 addition & 1 deletion zookeeper-server/src/main/java/org/apache/zookeeper/server/quorum/LearnerHandler.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -65,7 +65,7 @@ public class LearnerHandler extends ZooKeeperThread {

private static final Logger LOG = LoggerFactory.getLogger(LearnerHandler.class);

public static final String LEADER_CLOSE_SOCKET_ASYNC = "leader.closeSocketAsync";
public static final String LEADER_CLOSE_SOCKET_ASYNC = "zookeeper.leader.closeSocketAsync";
public static final boolean closeSocketAsync = Boolean.parseBoolean(System.getProperty(LEADER_CLOSE_SOCKET_ASYNC, "false"));

static {
Expand Down