what can I do to prevent downtime?

[Vick Khera]

I'm running wordpress on app engine flex. From time to time we see this failure on the browser:

Error: Server Error

The server encountered a temporary error and could not complete your request.
Please try again in 30 seconds.

I turned on the stackdriver monitoring for downtime to see if I could find out why. Since I turned it on last week, I've had four downtime alerts. The most recent was last night where we had two in succession. This was while I was having dinner, so nobody was manipulating the service other than visitors viewing the site.

Stackdriver has detected that one of your resources has recovered. 
Summary: The uptime check for morebusiness-com default has returned to a normal state. 
Violation Began: 2017-11-20 22:27:26 UTC (8 mins 54 secs) 
Condition Name: HTTP check on GAE Service default 
Violation Lasted: 5 mins 2 secs 

Is this normal? Are there tools that can help me figure out if I'm under-tuned for the VMs? I'm using the default App Engine Flex resource settings. My expectation was that App Engine would be more reliable than I'm observing.

This site makes pretty good money and the downtime is expensive for me. I was self-hosting for the last 20+ years and just moved to GAE two months ago to let someone else worry about being up all the time :)

Any advice will be appreciated. Thanks in advance.

[George (Cloud Platform Support)]

Your project's logs show quite a few errors for the Cloud HTTP Load Balancer. You may consider checking these log entries and investigating possible error origins in your app's code and load balancing settings. 

Related solutions have been provided on stackoverflow in reply to the "Compute Engine HTTP Load Balancing 502 errors" question and to "Google Cloud HTTP balancer returns 502 error". I encourage you to take a look at these 2 pages, and maybe adopt solutions suitable to your situation. If this still does not help, you are most welcome sharing the remaining issues here.

[Vick Khera]

On Tuesday, November 21, 2017 at 4:28:10 PM UTC-5, George (Cloud Platform Support) wrote:

Your project's logs show quite a few errors for the Cloud HTTP Load Balancer. You may consider checking these log entries and investigating possible error origins in your app's code and load balancing settings. 

Yes, that big batch of errors around 9am EST was due to me testing out the resources section of app.yaml, and not specifying everything in it. Apparently if you specify some parameters, the others do not get their defaults, and the app becomes unresponsive. Given how long it takes to re-deploy, I was down for over 10 minutes. :( Other than that there are only a handful of them.

Related solutions have been provided on stackoverflow in reply to the "Compute Engine HTTP Load Balancing 502 errors" question and to "Google Cloud HTTP balancer returns 502 error". I encourage you to take a look at these 2 pages, and maybe adopt solutions suitable to your situation. If this still does not help, you are most welcome sharing the remaining issues here.

 Thanks. I'll try to figure out why two instances may not be enough to handle the load and check back in here.

[George (Cloud Platform Support)]

Hi Vick, 

You are encouraged to post your investigation's results here, and the eventual fixes, for the benefit of the community at large.