-
Notifications
You must be signed in to change notification settings - Fork 1.3k
Description
We have a jar application
we are using the below "curator-framework-4.3.0" and "zookeeper-3.5.7" on the client servers for such application
CURATOR-525 There is a race condition in Curator which might lead to fake SUSPENDED event and ruin CuratorFrameworkImpl inner state - ASF JIRA (State Bug)
CURATOR-526 Error logged for valid config - "Invalid config event received: {properties}" - ASF JIRA (logging Bug)
Above two Bugs Fixed in 5.0.0/5.2.0
We have the below logs on our application (probably a combination of these two bugs) which ends up on consuming a lot of CPU and a failure after (due to open/close connections)
2024-10-18T06:35:27.855+11:00 - o.a.c.f.state.ConnectionStateManager - State change: SUSPENDED
2024-10-18T06:35:27.855+11:00 - - New connection state : SUSPENDED
2024-10-18T06:35:27.855+11:00 - - Connection to server has been suspended..
2024-10-18T06:35:28.638+11:00 - org.apache.curator.ConnectionState - Negotiated session timeout: 100002024-10-18T06:35:28.639+11:00 - o.a.c.f.state.ConnectionStateManager - State change: RECONNECTED
2024-10-18T06:35:28.639+11:00 - o.a.c.f.imps.CuratorFrameworkImpl - Clearing sleep for 0 operations
2024-10-18T06:35:28.639+11:00 - - New connection state : RECONNECTED
2024-10-18T06:35:28.639+11:00 - - Connection to server has been unsuspended...2024-10-18T06:35:28.640+11:00 - o.a.c.framework.imps.EnsembleTracker - New config event received: {}2024-10-18T06:35:28.640+11:00 - - Reconnect worker starting
2024-10-18T06:35:28.640+11:00 - - Attempting to reconnect to the Herd
2024-10-18T06:35:28.640+11:00 - - Waiting for connection to server
2024-10-18T06:35:28.640+11:00 - - Connected to server
2024-10-18T06:35:28.640+11:00 - - Reconnect worker completed2024-10-18T06:35:28.647+11:00 - o.a.c.framework.imps.EnsembleTracker - Invalid config event received: {}
Our Questions
1.For the logging Bug "Invalid config event received" , we need to upgrade the Curator as per https://issues.apache.org/jira/browse/CURATOR-526 please confirm?
2. For the reconnections , we are suspecting that it is related to BUG https://issues.apache.org/jira/browse/CURATOR-525 , however we need to confirm this (how?), also what is the trigger for this BUG knowing that we have another jar applications having the same Curator/Zookeeper version and they are not facing this BUG if our assumption is correct ? please advice the trigger here and why it is happening on some applications and not happening for others. also do we have any other solution rather than upgrading the curator version if this is the root cause?
3. Can we upgrade the curator to version 5.2.0 without upgrading the Zookeper knowing that we use Zookeeper "zookeeper-3.5.7"?