Skip to content
Browse files
[NO ISSUE][REP] Do not fail create/drop resource on replica failure
- user model changes: no
- storage format changes: no
- interface changes: no


- When creating/dropping a resource, do not fail the operation
  on the master node if it fails to send the opreation to the
  replica. When the replica node is recovered, it will be
  re-synced and will pick up all the changes.

Change-Id: I8313569cfda4b7e4fbeb23d4fdb998b3805367e1
Integration-Tests: Jenkins <>
Tested-by: Jenkins <>
Reviewed-by: Murtadha Hubail <>
Reviewed-by: Ali Alsuliman <>
  • Loading branch information
mhubail committed Aug 26, 2021
1 parent e7f6f72 commit 07d202d17ab0131703cc6731506d9880d180e65b
Showing 1 changed file with 8 additions and 2 deletions.
@@ -212,7 +212,11 @@ public synchronized void insert(LocalResource resource) throws HyracksDataExcept
resourceCache.put(resource.getPath(), resource);
//if replication enabled, send resource metadata info to remote nodes
if (isReplicationEnabled) {
createReplicationJob(ReplicationOperation.REPLICATE, resourceFile);
try {
createReplicationJob(ReplicationOperation.REPLICATE, resourceFile);
} catch (Exception e) {
LOGGER.error("failed to send resource file {} to replicas", resourceFile);

@@ -233,8 +237,10 @@ public synchronized void delete(String relativePath) throws HyracksDataException
FileReference resourceFile = getLocalResourceFileByName(ioManager, relativePath);
try {
if (resourceFile.getFile().exists()) {
if (isReplicationEnabled) {
try {
createReplicationJob(ReplicationOperation.DELETE, resourceFile);
} catch (Exception e) {
LOGGER.error("failed to delete resource file {} from replicas", resourceFile);
final LocalResource localResource = readLocalResource(resourceFile.getFile());

0 comments on commit 07d202d

Please sign in to comment.