Continuing the discussion from Identify: Taxon selected but cannot save:
Oh, I have this problem too, but I think I have a bit more insight into it. It seems to be that the search starts running on the first few letters typed in, but then stalls out. So in your example, “chor” seems to be the search term, despite the fact that you continued to type. When this happens to me, I delete letters until the search catches up and then add them back in. So for this, delete back to chor, then continue to type chortophaga when the search starts working again.
Yeah funny enough I recorded my screen just a few days ago when this was happening a lot, with intentions of submitting a bug report at some point.
I have a fast, stable internet connection so I know it’s not related to that.
@jwidness Sounds right, worked for me a few times. Does tack on extra time though. And yeah I don’t think it’s my internet connection.
I have the same thing quite regularly, and today I noticed it in the web uploader at well. I have never noticed it on the observation detail page though…
I get this often (Chrome on WIndows 7), and have the same issue with @usernames and when changing ‘Place’ for ‘Suggestion’ of the ‘Compare’ or ‘Identify’ pages. It can be very frustrating. When it happens, I wish I could stop iNat from offering the drop down list until I’ve clicked on a ‘go’ button.
This happens in all dropdowns - fields, projects, pinned locations. Also I’ll arrow down to my choice if it is in the list, and then the list reformats itself and either disappears altogether or reorders the items and I end up clicking on the wrong one.
And, even if I arrow down to my choice and click enter (or even if I click on it with the mouse), it chooses a different item. I can only choose the item I want (in some instances) if it is the only item in the list.
I haven’t been able to replicate.
Does this happen right when you start identifying, or does it usually crop up after you’ve gone through a few pages?
I’m rarely deeper than page 2 of Identify. It happens on page 1 for sure.
Stalling out on Berteroa incana, Erigeron strigosus, and Melilotus officinalis.
I also get this regularly when putting the place in in identify…
did this really just start happening lately? it seems like this must be an old issue.
i bet what’s happening is that you’re working faster than the system likes. i believe that each time you type (or remove) a character in that taxon selection box, you’re hitting the API with an autocomplete request, and i don’t think the system likes it when you hit it with too many requests over a short period of time. the API notes say you’re supposed to try to keep requests below 100 per minute, and my experience is that if you start exceeding a certain running average request rate (maybe something like 1 request per sec over the last 60 seconds), the API starts refusing your requests.
if you pull up the your browser’s developer tools and look at the network activity, you’ll see a request each time you type (or remove) a letter in that taxon box. try typing “bunnies” then backspacing then retyping, etc. a few times, and you should see that eventually the console starts logging errors to the effect of “too many requests”, and that’s the point where you’ll no longer get the proper autocomplete suggestions for bunnies (which can appear to be “stalling out” in some cases). if you wait a bit (thereby allowing your running average request rate to drop), the API will start accepting requests again. (so the next letter you type or remove from the box should return proper autocomplete results.)
i think this kind of thing could be what’s also causing the issue described here (https://forum.inaturalist.org/t/place-weirdness-in-identify-modal/4236), since the lookup of the country is also an API request.
UPDATE: i improved some of the text above, and just to help people understand what i was saying a little better, i’m attaching this screenshot below:
on the left side of the screenshot, you’ll see that I’ve forced a taxon selection box to “stall”. i typed “bunnies” into the box, but judging by the items in the taxa menu, the autocomplete appears to have “stalled” at “bu”. now look at the top right quadrant of the screenshot. you’ll see the last few items are autocomplete requests. you can see that starting with the autocomplete request for “bun”, i started getting 429 error codes. looking at the bottom right quadrant of the screenshot, you can see that 429 translates to “TOO MANY REQUESTS: the server is refusing to service the request because too many requests have been submitted by the client…” in other words, the last thing the server gave me successfully were autocomplete results for “bu”, and it refused to process anything else after that, which is why the taxon selection menu shows me things like “Buteos” and “Bugs” and “Butterflies” (not “Bunnies”).
This is definitely a new issue. It is making observation upload take way longer than it needs to. I think it is something to do with the Auto-ID process, as if I wait a long time before typing, it has no problems and works as normal.
If you enter the text and it hasn’t updated, it never will until you alter the text entered (deleting a letter, or adding one). This screenshot is after I left the screen for about 15 minutes.
I have been having this for a while, but only occasionally.
Thanks for the screenshot of the browser console showing the HTTP 429 errors. We do throttle some requests to prevent rogue API users from overloading the system. These rules change from time to time based on how people use (or abuse) the API. In the last couple weeks some additional rules were added and one rule inadvertently affected the taxa autocomplete endpoint, which is one we intentionally don’t want to throttle as much since it gets called a lot during a typical iNaturalist workflow. I’ve just fixed that issue, so hopefully people won’t see stalled taxon searches because of throttling anymore.
I’m curious to hear from anyone that’s been having this problem as they use the site over the next few days - does it feel like anything has changed with this problem since this fix just now? Has it gotten, worse, the same, better, or fixed entirely?
@pleary – thanks for addressing this. my initial impressions are that this is fixed. since you said this was fixed, i haven’t been able to force the taxon autocomplete to “stall” like i did before. i’ll point out that there were some mentions in this thread of other kinds of autocomplete that could potentially encounter this problem, though i didn’t try to reproduce any of those other potential problems. actually, i never encountered the taxa problem myself under my normal use because obviously i work much more slowly than some. so i’ll let the power users out there speak to whether the problem really appears to be fixed.
This does appear to be fixed. No issues in my latest batch. Fantastic.
Shall I mark as solved?
Looks like a similar issue is occurring in https://forum.inaturalist.org/t/loading-observations/4883