Lightsail: extreme slowness of both AWS interface and instances in Ohio (us-east-2) zone A


In the last three days, I've been encountering a cluster of problems that are making me rethink the viability of Lightsail as a platform.

On the 28th of September, I found three live WordPress instances I'm responsible for suffering heavy processor use for no apparent reason. Thinking this was a result of an attack of some kind, I checked on a dev instance mirroring the worst hit of these (an ecommerce site running on a 4-CPU instance), and found that it too was thrashed and had burned through its burst capacity, despite not even having a static IP address. At the same time, was very slow to fully load (I got to watch the little rocket tracing the infinity symbol for minutes at a time) and trying to reach the command line through that route was a lost cause. Yesterday was better, but today the interface slowness is back, and both public-facing and dev websites are timing out.

I see that there were problems on the Oregon side, but can't see a record of issues that should be affecting me. (If it's relevant, I'm located in Minnesota.)

Can anyone shed any light here?

  • Could you try to identify the processes that are causing the heavy CPU usage on your WordPress instances? That could provide some more hints as to what the issue is.

    About the console slowness, are you experiencing this slowness on the console in general, or only on specific pages?

asked 2 months ago32 views
1 Answer

Using the top command may help identify expensive processes exhausting CPU on your instance.

Given the issue manifests on both live site and dev sites that serve little traffic, it may point to an issue with your WordPress. With WordPress, many components can impact CPU usage such as: WordPress version, Plugins, Themes, WP-cron. You may want to identify when the CPU usage spiked and if it corresponds with installing or updating any of the above CP components.

You may also find cacheing or media offloading help alleviate CPU load. Offload and cacheing can be achieved through a variety of plugins, and possibly use of buckets or distributions. However, this sounds less likely as you mentioned the dev instance mirrors also being similarly impacted by CPU usage.

answered a month ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.

Guidelines for Answering Questions