I'm using Rising Wave in Docker with the 'playground' mode and the built-in postgres-cdc connector. After setting up tables and materialized views and letting the container idle without any database changes or queries, the memory usage levels off. However, after 15-20 minutes, the memory suddenly spikes, and the container crashes due to an Out of Memory (OOM) error. The logs don't show anything unusual before the spike. Is there a regularly scheduled process that could be causing this, and can I throttle it to prevent the OOM crash?
Rick Otten
Asked on Sep 19, 2023
The issue doesn't seem to be related to an idle timeout, as the playground mode in Docker has an unlimited idle timeout by default. It's more likely a slow memory leak or a background process that's causing the memory usage to grow until it runs out. While you're currently using the playground mode for an all-in-memory solution, it's not recommended for serious testing or production use. Rising Wave with persistent storage is designed to perform close to in-memory, with asynchronous disk writes and a well-optimized cache design. Even for large and complex joins, the computation resources are typically the bottleneck, not disk I/O. Therefore, a persistent store deployment is strongly recommended, even if you're looking for high performance and minimal disk overhead.