You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@kylin.apache.org by "powerinf (JIRA)" <ji...@apache.org> on 2018/06/06 06:33:00 UTC
[jira] [Commented] (KYLIN-2780) QueryController.getMetadata and
CacheController.wipeCache may be deadlock
[ https://issues.apache.org/jira/browse/KYLIN-2780?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16502858#comment-16502858 ]
powerinf commented on KYLIN-2780:
---------------------------------
saveCubeDesc and wipeCache can also make deadlock happenning ?
look the file kylin_deadlock.log
> QueryController.getMetadata and CacheController.wipeCache may be deadlock
> -------------------------------------------------------------------------
>
> Key: KYLIN-2780
> URL: https://issues.apache.org/jira/browse/KYLIN-2780
> Project: Kylin
> Issue Type: Bug
> Components: Query Engine
> Affects Versions: v1.5.4.1, v2.0.0
> Reporter: Fangyuan Deng
> Assignee: liyang
> Priority: Major
> Fix For: v2.1.0
>
> Attachments: KYLIN-2780.0.patch, jstack.log
>
>
> When we send update metadata request to the query sever manually,
> while some getMetadata requests come,
> deadlock may happen.
> The jstack log is listed below.
> and we add a double check in Broadcaster.getInstance to prevent lock each time.
> public static Broadcaster getInstance(KylinConfig config) {
> Broadcaster r = CACHE.get(config);
> if (r != null) {
> return r;
> }
> synchronized (CACHE) {
> r = CACHE.get(config);
> if (r != null) {
> return r;
> }
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)