[CASSANDRA-16535] Fail to init Cassandra Startup with an API connected with other cluster machines (3.11.9 version)

• Type: Bug
• Status: Resolved
• Priority: Normal
• Resolution: Won't Fix
• Fix Version/s: 3.11.11
• Component/s: Local/Startup and Shutdown
• Bug Category: Availability
• Severity: Normal
• Complexity: Low Hanging Fruit
• Discovered By: User Report
• Platform: All

Test and Documentation Plan

Specified in post

• Assignee: Aramis
• Reporter: Aramis

Authors

Aramis

Reviewers

Yuki Morishita

• Created: 24/Mar/21 20:14
• Updated: 23/Jul/21 15:36
• Resolved: 23/Jul/21 15:36

Description

The code in the function "VerifyPortsAreAvailable", look for the ports: 7000, 9042, 7199 and 9160 in the "netstat -an" command output.

 

If we have an API connected to the cluster in a node, and we try to start the execution of cassandra (with that API working, accesing ports 9042 of the other cluster machines), the output of that command return us that the port is already in use, and Cassandra´s startup process fails, showing the output message : "port already in use" in the .log file.

Thus, if we change the code of the function, adding the line: " -and $line -match "LISTENING" ", just next to the "$line -match "TCP" ", we will only interrupt the execution if the load port is in the state: "LISTENING", but no if it´s an active connection on the other node of the cluster.