Hi,
I’ve noticed that we get a large spike in HTTP response times every 5 minutes. I can’t think why this would be. The fly machines just read from a Redis database hosted on Redis Cloud, and I don’t see any latency spikes in Redis. We don’t have any cron jobs that run every 5 mins either or anything either. Don’t know if anyone has experienced anything similar or has any idea what might be happening?
Can you share your memory and CPU graph to see if that correlates with the spikes?
Is it worth deploying another type of machine, say an Apache instance, to see if you get a similar graph? This will help you discover if Redis is doing something (e.g. garbage collection) that could be causing this.
Hmmm just shooting from the hips here… is there any rate limiting on your redis?
Yeh will try this I guess. At the moment it’s just a simple node Fastify service with one endpoint that takes an id and fetches the object with the corresponding id from Redis. If it’s not Redis GC stuff or the like then not sure what else it could be other than Fly strangeness. Thing is I see no matching latency spikes in Redis. Note that we are currently dealing with an issue where with CPU saturation in Redis that is causing latency issues (but not spikey in the way we see in Fly). Screenshot is of Redis metrics.
hmmm not that I know of or can see
This topic was automatically closed 7 days after the last reply. New replies are no longer allowed.
