[
https://issues.apache.org/jira/browse/HADOOP-17358?focusedWorklogId=510010&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-510010
]
ASF GitHub Bot logged work on HADOOP-17358:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 10/Nov/20 23:56
Start Date: 10/Nov/20 23:56
Worklog Time Spent: 10m
Work Description: amahussein commented on a change in pull request #2436:
URL: https://github.com/apache/hadoop/pull/2436#discussion_r520949563
##########
File path:
hadoop-common-project/hadoop-common/src/main/java/org/apache/hadoop/conf/Configuration.java
##########
@@ -2876,12 +2876,28 @@ public Reader getConfResourceAsReader(String name) {
protected synchronized Properties getProps() {
if (properties == null) {
properties = new Properties();
- Map<String, String[]> backup = updatingResource != null ?
- new ConcurrentHashMap<String, String[]>(updatingResource) : null;
- loadResources(properties, resources, quietmode);
+ loadProps(properties, 0, true);
+ }
+ return properties;
+ }
+ /**
+ * Loads the resource at a given index into the properties.
+ * @param props the object containing the loaded properties.
+ * @param startIdx the index where the new resource has been added.
+ * @param fullReload flag whether we do complete reload of the conf instead
+ * of just loading the new resource.
+ * @return the properties loaded from the resource.
Review comment:
Sorry . I overlooked that.
I added a new commit.
----------------------------------------------------------------
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.
For queries about this service, please contact Infrastructure at:
[email protected]
Issue Time Tracking
-------------------
Worklog Id: (was: 510010)
Time Spent: 1h 50m (was: 1h 40m)
> Improve excessive reloading of Configurations
> ---------------------------------------------
>
> Key: HADOOP-17358
> URL: https://issues.apache.org/jira/browse/HADOOP-17358
> Project: Hadoop Common
> Issue Type: Bug
> Components: conf
> Reporter: Ahmed Hussein
> Assignee: Ahmed Hussein
> Priority: Major
> Labels: pull-request-available
> Time Spent: 1h 50m
> Remaining Estimate: 0h
>
> [~daryn] reported that adding a new resource to a conf forces a complete
> reload of the conf instead of just loading the new resource. Instantiating a
> {{SSLFactory}} adds a new resource for the ssl client/server file. Formerly
> only the KMS client used the SSLFactory but now TLS/RPC uses it too.
> The reload is so costly that RM token cancellation falls behind by hours or
> days. The accumulation of uncancelled tokens in the KMS rose from a few
> thousand to hundreds of thousands which risks ZK scalability issues causing a
> KMS outage.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
