Magnolia author or public slow response
Symptom
A CustomerMagnoliaAuthorSlowResponse
alert is firing or a CustomerMagnoliaPublicSlowResponse
alert is firing. Both alerts use the same
metrics but with different thresholds:
-
The author instance is slow responding if the average response time for 90% of requests is more than 2 seconds
-
A public instance is slow responding if the average response time for 90% of requests is more than 0.8 seconds
CustomerMagnoliaAuthorSlowResponse and CustomerMagnoliaPublicSlowResponse alerts are sent to subscribers via email.
|
Observations
Solutions
This section provides solutions that should help resolve the issue in most cases.
Investigate slow response cause
Slow responses from Magnolia can be caused by many things, including the implementation of the your project.
Here’s an incomplete list of possible causes of slow response times and what actions to take:
Cause | Action |
---|---|
High request volume |
Check the rate of requests to Magnolia. |
Insufficient memory |
Check the memory allocation for the Magnolia pod and the JVM max heap setting. |
Excessive garbage collection |
Check for frequent garbage collection and how close and how frequently the JVM heap used is to the JVM max heap. |
Magnolia cache is underutilized |
Check the Magnolia cache hit percentage. |
Custom filter bottleneck |
Check for custom filters in the filter chain. |
Slow requests per type |
Turn on detailed response time metrics and check for slow requests by type (e.g., |
Check the miss request rate, origin latency, and miss latency dashboards in Fastly. These dashboards often indicate if Magnolia is always slow or sometimes slow. |