Bug 32623031 - NDB_BACKUP_NODEFAIL FAILS ON PB2

Maitrayi Sabaratnam · Maitrayi Sabaratnam · commit 8998a98f93aa · 2021-03-15T19:49:26.000+01:00
Followup fix for BUG#32381165 - MYSQL-8.0 (8.0.24): NODE CRASHED
DURING BACKUP WITH ERROR "SIGNAL 11 RECEIVED

Problem: A node restart completion is informed only to the worker-0
handling backup-thread-0 and not to the other thread workers.

Fix: Inform all backup threads that a node has restarted.

Reviewed by: Frazer Clement &lt;frazer.clement@oracle.com&gt;
diff --git a/storage/ndb/src/kernel/blocks/backup/BackupProxy.cpp b/storage/ndb/src/kernel/blocks/backup/BackupProxy.cpp
@@ -218,9 +218,12 @@ BackupProxy::sendSUM_EVENT_REP(Signal* signal, Uint32 ssId)
 void
 BackupProxy::execNODE_START_REP(Signal *signal)
 {
-    jam();
-    sendSignal(workerRef(0), GSN_NODE_START_REP, signal,
+  jam();
+  for (Uint32 i = 0; i < c_workers; i++)
+  {
+    sendSignal(workerRef(i), GSN_NODE_START_REP, signal,
                signal->getLength(), JBB);
+  }
 }
 
 

Original file line number	Diff line number	Diff line change
`@@ -218,9 +218,12 @@ BackupProxy::sendSUM_EVENT_REP(Signal* signal, Uint32 ssId)`
`218`	`218`	`void`
`219`	`219`	`BackupProxy::execNODE_START_REP(Signal *signal)`
`220`	`220`	`{`
`221`		`- jam();`
`222`		`- sendSignal(workerRef(0), GSN_NODE_START_REP, signal,`
	`221`	`+ jam();`
	`222`	`+ for (Uint32 i = 0; i < c_workers; i++)`
	`223`	`+ {`
	`224`	`+ sendSignal(workerRef(i), GSN_NODE_START_REP, signal,`
`223`	`225`	`signal->getLength(), JBB);`
	`226`	`+ }`
`224`	`227`	`}`
`225`	`228`
`226`	`229`