devmadhuu commented on code in PR #4696:
URL: https://github.com/apache/ozone/pull/4696#discussion_r1192054732
##########
hadoop-ozone/recon/src/main/java/org/apache/hadoop/ozone/recon/tasks/ContainerKeyMapperTask.java:
##########
@@ -383,4 +398,60 @@ private void handlePutOMKeyEvent(String key, OmKeyInfo
omKeyInfo,
}
}
+ /**
+ * Write an OM key to container DB and update containerID -> no. of keys
+ * count to the Global Stats table.
+ *
+ * @param key key String
+ * @param omKeyInfo omKeyInfo value
+ * @param containerKeyCountMap we keep the containerKey counts in this map
+ * to allow batching to containerKeyCountTable
+ * after reprocessing is done
+ * @throws IOException if unable to write to recon DB.
+ */
+ private void handleKeyReprocess(String key,
+ OmKeyInfo omKeyInfo,
+ Map<Long, Long> containerKeyCountMap)
+ throws IOException {
+ long containerCountToIncrement = 0;
+ for (OmKeyLocationInfoGroup omKeyLocationInfoGroup : omKeyInfo
+ .getKeyLocationVersions()) {
+ long keyVersion = omKeyLocationInfoGroup.getVersion();
+ for (OmKeyLocationInfo omKeyLocationInfo : omKeyLocationInfoGroup
+ .getLocationList()) {
+ long containerId = omKeyLocationInfo.getContainerID();
+ ContainerKeyPrefix containerKeyPrefix = new ContainerKeyPrefix(
+ containerId, key, keyVersion);
+ if (reconContainerMetadataManager.getCountForContainerKeyPrefix(
+ containerKeyPrefix) == 0) {
+ // Save on writes. No need to save same container-key prefix
+ // mapping again.
+ reconContainerMetadataManager.storeContainerKeyMapping(
Review Comment:
Thanks for the patch @ivandika3 , though I have one suggestion. After this
change, we are storing containerKeyMapping info with every key which is hitting
DB everytime, so if we want to improve memory footprint due to containerKeyMap,
we can use a threshold and till that threshold number of keys, retain in
memory, lets say 1000 keys, and then use batchStore function for
containerKeyMap instead of removing it completely. We have done like this in
other aync tasks like NSSummary, You can refer that. This way we'll avoid 1
billion DB calls with 1 billion keys.
--
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]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
