Skip to content

Fix bug that bad replica can not be synchronized when report #1634

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

Merged
chaoyli merged 1 commit into from
Aug 14, 2019
Merged

Fix bug that bad replica can not be synchronized when report #1634

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
1 change: 1 addition & 0 deletions fe/src/main/java/org/apache/doris/catalog/Catalog.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4227,6 +4227,7 @@ private void unprotectUpdateReplica(ReplicaPersistInfo info) {
Replica replica = tablet.getReplicaByBackendId(info.getBackendId());
Preconditions.checkNotNull(replica, info);
replica.updateVersionInfo(info.getVersion(), info.getVersionHash(), info.getDataSize(), info.getRowCount());
replica.setBad(false);
}

public void replayAddReplica(ReplicaPersistInfo info) {
Expand Down
10 changes: 8 additions & 2 deletions fe/src/main/java/org/apache/doris/catalog/TabletInvertedIndex.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -349,11 +349,17 @@ private boolean needSync(Replica replicaInFe, TTabletInfo backendTabletInfo) {
}
long versionInFe = replicaInFe.getVersion();
long versionHashInFe = replicaInFe.getVersionHash();

if (backendTabletInfo.getVersion() > versionInFe
|| (versionInFe == backendTabletInfo.getVersion()
&& versionHashInFe != backendTabletInfo.getVersion_hash())) {
|| (versionInFe == backendTabletInfo.getVersion() && versionHashInFe != backendTabletInfo.getVersion_hash())) {
// backend replica's version is larger or newer than replica in FE, sync it.
return true;
} else if (versionInFe == backendTabletInfo.getVersion() && versionHashInFe == backendTabletInfo.getVersion_hash()
&& replicaInFe.isBad()) {
// backend replica's version is equal to replica in FE, but replica in FE is bad, while backend replica is good, sync it
return true;
}

return false;
}

Expand Down
6 changes: 4 additions & 2 deletions fe/src/main/java/org/apache/doris/master/ReportHandler.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -436,7 +436,8 @@ private static void sync(Map<Long, TTablet> backendTablets, ListMultimap<Long, L
}

if (metaVersion < backendVersion
|| (metaVersion == backendVersion && metaVersionHash != backendVersionHash)) {
|| (metaVersion == backendVersion && metaVersionHash != backendVersionHash)
|| (metaVersion == backendVersion && metaVersionHash == backendVersionHash && replica.isBad())) {

// This is just a optimization for the old compatibility
// The init version in FE is (1-0), in BE is (2-0)
Expand All @@ -455,7 +456,8 @@ private static void sync(Map<Long, TTablet> backendTablets, ListMultimap<Long, L
// 1. PUSH finished in BE but failed or not yet report to FE
// 2. repair for VERSION_INCOMPLETE finished in BE, but failed or not yet report to FE
replica.updateVersionInfo(backendVersion, backendVersionHash, dataSize, rowCount);

replica.setBad(false);

if (replica.getLastFailedVersion() < 0 && !isInitVersion) {
// last failed version < 0 means this replica becomes health after sync,
// so we write an edit log to sync this operation
Expand Down