This article provides instructions on how to collect data and logs manually for support to analyze an issue if OpsCenter is unavailable.
Support requires access to a number of files and logs to troubleshoot issues. If OpsCenter is unavailable or a diagnostic file cannot be generated from it, the following steps allow to collect the necessary information to investigate issues.
There is 2 ways to collect data when OpsCenter isn't available. The first solution shows which logs and output to collect manually. The second provides an external shell script allowing for the process to be done automatically on each node.
Note that if OpsCenter is available, the easiest way to go forward is to follow instructions provided here:
Option 1 - Manual log collection
Please collect the following from all nodes and place the outputs/files in a directory with the node’s IP address as the directory name:
debug.logif using DSE 5.x
- from each node, the output of:
$ nodetool status > status
$ nodetool info > info
$ nodetool cfstats > cfstats
$ nodetool compactionstats > compactionstats
$ nodetool describecluster > describecluster
$ nodetool gossipinfo > gossipinfo
$ nodetool proxyhistograms > proxyhistograms
$ nodetool netstats > netstats
$ nodetool tpstats > tpstats
$ dsetool ring > dsetool-ring
$ dsetool status > dsetool-status
$ java -version > java-version.txt
$ free -m > free-m.txt
$ df -h > df-h.txt
From one node only:
cqlsh -e "describe full schema" > schema.full
Option 2 - Scripted log collection
DISCLAIMER: The following links are provided as is by engineers from DataStax Support to ease the log collection process but do not fall under DataStax official offering. The scripts have not undergone full QA testing and should be run in a test system before running them in a production environment.
Download and run the script provided in the link below on all the nodes: