On Mobile, tree details continuously refreshes or goes back to map.

[zoobot]

Check on actual phone

[zoobot]

@mwpark2014 Could this is related to commit 8b400b81b98dd107bd437c9a8b2757829601cbd7

Does it happen on your phone?

[mwpark2014]

Ooh, sorry if this is me. I won't be able to fix until late tomorrow.

[mwpark2014]

@zoobot I checked on my phone though. I'm not sure what the issue you are seeing is. For me, I don't see "continuously refreshes or goes back to map", but I run into a "Something went wrong while displaying this webpage" after staying on the tree details for a while, which is super weird

I can only repro this on my phone's chrome browser on android. Not on the mobile emulator in the chrome browser

[zoobot]

Aww frustrating! I can't replicate it using dev tools on desktop. Maybe it has something to do with touch somewhere.

I'll look around at the Commits for changes to that as well.

[zoobot]

@mwpark2014 On dev I see "An error occurred. Please see console for more information" on my actual phone. I am going to try and go back through the commits until it stops failing.

[zoobot]

These ones seem fine

• dcf4d6c - Update default stale time to 5 min (5 weeks ago)
• 8b400b8 - Replace useEffect handling tree selection in MapLayout (#410) (5 weeks ago)
• e31ba2b - Update README.md (#421) (5 weeks ago)

This one causes a refresh but don't see an error.

• 0ffb5da - Get rid of updating query cache Tree vector data and tree db data is already being combined correctly in MapLayout (5 weeks ago)

[zoobot]

I tried reverting and it breaks onclick of the points.

[mwpark2014]

Check on actual phone

Can you share what phone and what browser you're using and steps to repro?

[zoobot]

iphone 12 & 5se
When scrolling to the bottom of the tree, it refreshes. Also the map refreshes randomly.
Sometimes it refreshes from the tree and goes back to the map.

[mwpark2014]

This was first discovered yesterday, right? There was no indicator of this bug before? I connected my android phone using chrome browser to a local machine and I am able to repro the issue

Using https://waterthetrees.com/#pos=15/37.7207/-122.43923&id=4469999575224304
The local machine developer tools identified "potential out-of-memory crash", which explains why we don't see this on desktop. It takes 16 seconds to crash on my device. I wonder if it takes less time on less powerful devices.

Taking a deeper look into the code where this comes up

Update: Took a deeper look into the code. It doesn't look to be part of our codebase. I'm fairly certain that this is just general mapbox code that maps out points. Instead, let's focus on investigating what changes could have caused this issue to appear now

[mwpark2014]

Reproed the huge memory usage on desktop

Memory usage for this one tab (I'm 90% sure this is the wtt tab) spikes up and peaks at 2gb, then floats down to 1.5gb and settles. The pattern seems strange cause it's not a memory leak in that it's continuously leaking memory. Also need to be open to other potential sources of this bug including mapbox. Doesn't seem to be browser specific though

Going to test if it's related to the tree details container in particular and on desktop to isolate the issue.

[mwpark2014]

Confirmed that this problem only occurs when the tree details container is open. It doesn't occur on the map without opening anything.
This happens on desktop and mobile even without the tree details container being open

Confirm that this memory issue exists on desktop too, but unlikely that desktop will crash due to higher RAM.

🤔 🤔 🤔

[mwpark2014]

I think this is a back-end change. Do you know what causes these two loading phases @zoobot? I know this wasn't a thing before

Phase 1:

Phase 2:

Confirmed that the phase 2 of loading corresponds with the memory overload

[zoobot]

I think you are right, wtt_server is spiking on load

[mwpark2014]

Okay, cool. It's getting late, so I'll be hands off for now. I can be available late tomorrow night to help out if necessary.

Thanks for finding the issue! I've been thinking that we need to set up monitoring to be aware of these types of issues in the future more quickly automatically. Maybe set up some alarms. It's on my list but was a little further down in priority

[zoobot]

Thanks for your sleuthing on this!! I think you found the issue!

[zoobot]

Temporarily fixes the cpu spike.

Need to deal with the tree-sources repo picking up all the trees modified in the db when the city is NOT in sources.
Better fix will be to add city and source at the same time as adding the first tree in a city.