CPU starvation with RDB loader version 5.5.0

Hey everyone, we just upgraded to RDB loader version 5.5.0 and we’re getting this error now. Is it related to the new update or is it something we could potentially resolve on our end?

INFO com.snowplowanalytics.snowplow.rdbloader: RDB Loader 5.5.0 has started.	sp_rdb_loader without the blocking combinator.
[WARNING] Your app's responsiveness to a new asynchronous event (such as a	sp_rdb_loader new connection, an upstream response, or a timer) was in excess of 100 milliseconds.
Your CPU is probably starving. Consider increasing the granularity	sp_rdb_loader of your delays or adding more cedes. This may also be a sign that you are	sp_rdb_loader unintentionally running blocking I/O operations (such as File or InetAddress)

Thanks a lot for your amazing work and responsiveness!

Hi @Kristina_Pianykh ,

The short answer is that you don’t need to worry about it.

That error message comes from one of the underlying 3rd party libraries that we used. (There’s a long technical article about it here – you don’t need to read it!).

It’s happening because we tend to run RDB loader with minimal cpu resources, e.g. 1 available cpu. This is a good strategy, because RDB loader does not do much “work”, it just coordinates running SQL statements in the warehouse. But it seems that by running with such little CPU we occasionally hit a bottleneck which makes this 3rd party library print a warning message.

We (Snowplowers) should aim to remove this warning from the logs in a future release, because it only causes confusion for Snowplow users. I don’t think you should change your configuration in any way.

2 Likes

Many thanks for your comprehensive explanation, much appreciated!