Below are several of our common issues we see in Support and what you can proactively provide to help Redpanda Support more efficiently review your ticket.
Console Issues
- Capture the problem with a screenshot or a quick video
- Also capture the problem in HAR format. See this KB for more information.
- Send the Console log. (Self Hosted Customers Only)
Client Issues
- Tell us the client implementation along with its version
- Send the full client log that shows the problem
Broker Crashing Issues (Self Hosted Customers Only)
- Send the log that covers the time the process crashed.
- Tell us about the redpanda version and the CPU architecture type.
Tiered Storage Issues
- Collect "rpk debug bundle --partition <comma separated problematic partitions list>" from all nodes
- Provide 2 or 3 partitions as a sample
Latency Issues
- Tell us how the latency is measured.
- Tell us how it looked before and how it looks now.
- Tell us what is the expected latency.
- Tell us about the deployment details; cloud vendors, machine type, disk details. (Self Hosted Customers Only)
- Tell us about any environment details, client and cluster, that have changed around the time of the change in latency.
Consumer Lag issues
- Tell us the consumer client implementation along with its version.
- Tell us the group & topic names in question.
- Tell us when the lag started to happen.
- Share some metrics charts that denote lags.
- Did the lag abruptly increase or gradually?
- If the lag abruptly increased, run "rpk debug bundle --logs-since xxx --logs-until yyy" and ensure it captures the time the lag happened. Also send all content under /var/lib/redpanda/data/kafka/__consumer_offsets from all nodes. (Self Hosted Customers Only)
- If the lag gradually increased, send the client logs along with "rpk debug bundle --logs-since xxx" (Self Hosted Customers Only)
Cluster Health Issue (Self Hosted Customers Only)
Leaderless Partitions
- Run "rpk cluster health" five times every 30 seconds and send all the outputs along.
- Collect "rpk debug bundle --partition <comma separated some problematic partition list>" from all nodes.
Under-replicated partitions
- Run "rpk cluster health" five times every 30 seconds and send all the outputs along.
- Collect "rpk debug bundle --partition <comma separated some problematic partition list>" from all nodes.