Plant Tracer instance appears to "hang" sometimes

[sbarber2]

This isn't easily reproducible, but occasionally when navigating to https://dev.planttracer.com, it appear that no response comes back. I wait 10-60 seconds depending on how patient I'm feeling, and all I see is completely white, blank frame.

Then I open another browser tab and issue the same request, and the response comes back immediately and I see the expected "Welcome to the Plant Tracer Demo!" page.

It's disconcerting to say the least to a new user, who will likely immediately assume the site is broken and we've already lost their confidence.

We've discussed this a bit but not resolved it; I believe the diagnosis was something like "Well, that's Lambda for you" but that doesn't help the new user problem.

For the Nov2024 milestone, the issue just needs to be documented with the workaround of hitting the site again.

But also, we should "fix" this so it just doesn't happen, or at least doesn't appear to the user to happen.

[simsong]

I don’t think it’slambda. I think that it may be trying to open the connection to DreamHost. You would have to look at the log files to see where the delay is.

…

On Sat, Oct 26, 2024 at 4:39 PM Steve Barber ***@***.***> wrote: This isn't easily reproducible, but occasionally when navigating to https://dev.planttracer.com, it appear that no response comes back. I wait 10-60 seconds depending on how patient I'm feeling, and all I see is completely white, blank frame. Then I open another browser tab and issue the same request, and the response comes back immediately and I see the expected "Welcome to the Plant Tracer Demo!" page. It's disconcerting to say the least to a new user, who will likely immediately assume the site is broken and we've already lost their confidence. We've discussed this a bit but not resolved it; I believe the diagnosis was something like "Well, that's Lambda for you" but that doesn't help the new user problem. For the Nov2024 milestone, the issue just needs to be documented with the workaround of hitting the site again. But also, we should "fix" this so it just doesn't happen, or at least doesn't appear to the user to happen. — Reply to this email directly, view it on GitHub <#615>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAMFHLHVOU4YEBRIZZX56A3Z5OZSBAVCNFSM6AAAAABQU46RESVHI2DSMVQWIX3LMV43ASLTON2WKOZSGYYTKOJUG43DIOI> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[sbarber2]

Ah, thanks. I don't believe I have access to the AWS log files. I think that is pending the work of #616

I should be able to see the Dreamhost logs, so will check them. For the record I saw this behavior sometime between 0930-1000 EDT on Fri Oct 25 2024, but without correlating them against the AWS logs they might not be that informative.

[simsong]

I'll try to figure out how to give our service account access to the logs.

[simsong]

I edited the title as we don't know that it's when an instance starts. Instance starts are actually incredibly fast. However, we could add to the footer the current instance ID and how long it has been running. You would be surprised by the results. (Typically there are 3-4 instances running at a time.)