sanpwc commented on code in PR #4043:
URL: https://github.com/apache/ignite-3/pull/4043#discussion_r1665802488
##########
modules/partition-replicator/src/main/java/org/apache/ignite/internal/partition/replicator/PartitionReplicaLifecycleManager.java:
##########
@@ -203,22 +225,115 @@ public CompletableFuture<Void>
startAsync(ComponentContext componentContext) {
inBusyLock(busyLock, () ->
onCreateZone(parameters).thenApply((ignored) -> false))
);
- return nullCompletedFuture();
+ return allOf(processZonesOnStart, processAssignmentsFuture);
+ }
+
+ private CompletableFuture<Void> processZonesOnStart(long recoveryRevision,
@Nullable HybridTimestamp lwm) {
+ int earliestCatalogVersion =
catalogMgr.activeCatalogVersion(hybridTimestampToLong(lwm));
+ int latestCatalogVersion = catalogMgr.latestCatalogVersion();
+
+ var startedZones = new IntOpenHashSet();
+ var startZoneFutures = new ArrayList<CompletableFuture<?>>();
+
+ for (int ver = latestCatalogVersion; ver >= earliestCatalogVersion;
ver--) {
+ int ver0 = ver;
+ catalogMgr.zones(ver).stream()
+ .filter(tbl -> startedZones.add(tbl.id()))
+ .forEach(zoneDescriptor -> startZoneFutures.add(
+
calculateZoneAssignmentsAndCreateReplicationNodes(recoveryRevision, ver0,
zoneDescriptor)));
+ }
+
+ return allOf(startZoneFutures.toArray(CompletableFuture[]::new))
+ .whenComplete((unused, throwable) -> {
+ if (throwable != null) {
+ LOG.error("Error starting zones", throwable);
+ } else {
+ LOG.debug("Zones started successfully");
+ }
+ });
+ }
+
+ private CompletableFuture<Void> processAssignmentsOnRecovery(long
recoveryRevision) {
+ var stableAssignmentsPrefix = new ByteArray(STABLE_ASSIGNMENTS_PREFIX);
+ var pendingAssignmentsPrefix = new
ByteArray(PENDING_ASSIGNMENTS_PREFIX);
+
+ CompletableFuture<Void> stableFuture = handleAssignmentsOnRecovery(
Review Comment:
It's tricky why we need to handle stableAssignments on recovery in addition
to handling stable within processZonesOnStart. Could you please add a comment
here and a common javadoc for processAssignmentsOnRecovery explaining that
stable handling here means destroy resources?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
