[ https://issues.apache.org/jira/browse/ZOOKEEPER-1925?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14633376#comment-14633376 ]
Ivan Balashov commented on ZOOKEEPER-1925: ------------------------------------------ Any reason TTL _should not_ be added to ZK? Here are only some problems with app-level cleaning: – Slow for large trees. If the network latency is high it takes a very long time for the app to walk whole tree (e.g. 1M nodes). – Risky. If the cleanup application crash (or becomes unreachable) it puts whole ZK cluster to danger. – The implementation might be error-prone if multiple cleanup threads are used. – Has to be implemented separately by each developer (until a proven implementation shared as a library). > Time to Live or auto expiration of zookeeper node > ------------------------------------------------- > > Key: ZOOKEEPER-1925 > URL: https://issues.apache.org/jira/browse/ZOOKEEPER-1925 > Project: ZooKeeper > Issue Type: New Feature > Reporter: Omkar Vinit Joshi > > Today whenever a znode is created; it stays there for ever. We all know that > there is a limitation in terms of how many nodes a system can handle. It > would be nice to have a way to specify expiry time for every znode thereby > stale zondes are cleandup automatically after sufficiently large time. any > thoughts? -- This message was sent by Atlassian JIRA (v6.3.4#6332)