HCL Commerce Version 9.1.5.0

Troubleshooting: ConnectionLoss error during first full index build

After the first full index build, the data-query service loses its connection to ZooKeeper. Characteristic error messages are written to the logs which allow you to diagnose and resolve this problem.

Problem

After first full index build, the data-query service generates a ConnectionLoss error associated with ZooKeeper. The result of this failure is that rule-based sales category operations and Search do not work. For example, if you have enabled rule-based sales categories in the Transaction server and have created such a category, selecting and reloading that category returns no data.

The ZooKeeper log contains a message similar to the following:
2021-01-12 19:31:33,022 [myid:1] 
- WARN [NIOServerCxnFactory.AcceptThread:0.0.0.0/0.0.0.0:2181:RateLogger@56] 
- [29 times] 
Message: Error accepting new connection: Too many connections from /172.20.0.6 - max is 60
In the data-query service log (by default /app/ESQueryService/logs/trace.log), there are errors similar to the following:
[2021-01-12 19:31:32,571] 
[Default Executor-thread-11] [6916958446773453208] 
ERROR c.h.c.s.i.c.SearchConfigurationRegistry.loadProfile:246 
- KeeperErrorCode = Session expired for /profiles
[2021-01-12 19:31:39,242] 
[Default Executor-thread-9] [6916958469548975730] 
ERROR c.h.c.s.i.c.SearchConfigurationRegistry.loadProfile:246 
- KeeperErrorCode = ConnectionLoss for /profiles
Where 172.20.0.6 is the data-query service address.

Solution

Restart the data-query service to reset the connection.