[enhance](mtmv)MTMV no longer generate cache when replaying logs (apa…

…che#44283)

### What problem does this PR solve?

When replaying logs, cache will no longer be generated because the
catalog may not have been initialized or the dependent environment may
not be connected, causing it to freeze here and preventing FE from
starting

The cost is that after the materialized view is refreshed, the cache of
the follower node will be empty, and a cache will be generated when the
query is first used

Problem Summary:
MTMV no longer generate cache when replaying logs

fe/fe-core/src/main/java/org/apache/doris/alter/Alter.java

@@ -1022,7 +1022,8 @@ public void processAlterMTMV(AlterMTMV alterMTMV, boolean isReplay) {
                     mtmv.alterMvProperties(alterMTMV.getMvProperties());
                     break;
                 case ADD_TASK:
-                    mtmv.addTaskResult(alterMTMV.getTask(), alterMTMV.getRelation(), alterMTMV.getPartitionSnapshots());
+                    mtmv.addTaskResult(alterMTMV.getTask(), alterMTMV.getRelation(), alterMTMV.getPartitionSnapshots(),
+                            isReplay);
                     break;
                 default:
                     throw new RuntimeException("Unknown type value: " + alterMTMV.getOpType());

fe/fe-core/src/main/java/org/apache/doris/catalog/MTMV.java

@@ -190,15 +190,19 @@ public MTMVStatus alterStatus(MTMVStatus newStatus) {
     }
 
     public void addTaskResult(MTMVTask task, MTMVRelation relation,
-            Map<String, MTMVRefreshPartitionSnapshot> partitionSnapshots) {
+            Map<String, MTMVRefreshPartitionSnapshot> partitionSnapshots, boolean isReplay) {
         MTMVCache mtmvCache = null;
         boolean needUpdateCache = false;
         if (task.getStatus() == TaskStatus.SUCCESS && !Env.isCheckpointThread()
                 && !Config.enable_check_compatibility_mode) {
             needUpdateCache = true;
             try {
-                // shouldn't do this while holding mvWriteLock
-                mtmvCache = MTMVCache.from(this, MTMVPlanUtil.createMTMVContext(this), true);
+                // The replay thread may not have initialized the catalog yet to avoid getting stuck due
+                // to connection issues such as S3, so it is directly set to null
+                if (!isReplay) {
+                    // shouldn't do this while holding mvWriteLock
+                    mtmvCache = MTMVCache.from(this, MTMVPlanUtil.createMTMVContext(this), true);
+                }
             } catch (Throwable e) {
                 mtmvCache = null;
                 LOG.warn("generate cache failed", e);