The Vertica node/nodes does/do not start.
Message in the vertica.log file:
2025-04-16 13:14:15.000 Cluster Inviter:0x7f9cba7fc700 [Comms] <INFO> I am not a member of JOIN group. Yet. Wait.
The following errors are displayed when launching the Vertica manually:
SP_connect: unable to connect via UNIX socket to /opt/vertica/spread/tmp/4803 (pid=1974127): Error: No such file or directory
SP_connect: unable to connect via UNIX socket to /opt/vertica/spread/tmp/4803 (pid=2082254): Error: Connection refused
DX NetOps PM 23.3.3
Vertica 10.1.1.20
The file descriptor fd=3 is being opened during Vertica startup, possibly by a library performing logging or auditing. Perhaps the kernel upgrade or OS patching introduced the issue.
The data repository dr_install.sh installation script upgrades Vertica from 10.1.1-20 to 11.1.1, then from 11.1.1 to 12.0.4, and finally, from 12.0.4 to 23.4.0. At each version upgrade, the script starts and stops the database.
Starting with version 11.1, Vertica properly handles the Spread descriptor during auto-restart scenarios, which prevents it from being confused by such descriptors that may be opened at startup.
Upgrade PM to the latest release which will also upgrade Vertica.
Test the UDP port 4803:
a) On one node (node 1) set the netcat in server mode for UDP port 4803
nc -ul 4803
nc -ul -p 4803
b) On the other node run the netcat in client mode for UDP port 4803 on first node
nc -u 10.152.44.198 4803
Then type any word and press Enter. The test message should be displayed on the netcat in server mode (first node)
1. Ensure the vertica is not running on any node.
pidof vertica
2. If vertica is running, kill it:
pkill -9 vertica