These are chat archives for arenanet/api-cdi

18th
Mar 2017
:fire: turned off the APIs :fire:
not sure what happened but the JS contexts have been crashing for 30 minutes
and my housemate was "hey JS contexts are exploding"
and not gonna lie
it's friday night and st patricks day and i'm wasted
I'll see if I can fix it in the morning :<
Eearslya Sleiarion
@Eearslya
Mar 18 2017 07:09
:fire:
oh boy
yeahhhhhh
darthmaim
@darthmaim
Mar 18 2017 11:12
don't worry, you got this :heart:
ChieftainAlex
@ChieftainAlex
Mar 18 2017 13:05
60969.png
anyone know if those icons have separate files in the .dat as well?
darthmaim
@darthmaim
Mar 18 2017 13:06
IIRC yes
ChieftainAlex
@ChieftainAlex
Mar 18 2017 13:06
I was expecting 32px
naturally my first port of call would have been the API but it's not a great day for that :D
darthmaim
@darthmaim
Mar 18 2017 13:08
64px (102320, ...)
ChieftainAlex
@ChieftainAlex
Mar 18 2017 13:08
excellent thank you, I must have scrolled past them :/
Archomeda
@Archomeda
Mar 18 2017 14:15
hmm... this probably means that my bot is broken
a good moment to fix the error message for next time <3
Archomeda
@Archomeda
Mar 18 2017 14:33
Discord_2017-03-18_15-33-07.png
:D
Archomeda
@Archomeda
Mar 18 2017 15:16
aw, the forum doesn't provide rss feeds for posts of individual users :(
John Carroll
@johncarroll
Mar 18 2017 17:10
hey all. does anyone have a cached accounts/achievements response they wouldn't mind sharing with me?
(a working one)
Darqam
@Darqam
Mar 18 2017 17:11
wiki should have 2-3 examples if you just want the structure of it
John Carroll
@johncarroll
Mar 18 2017 17:15
thanks, but i'm looking for something that has specific achievement id's and i didn't want to do busy work :D
i'll just wait if no-one has any
Archomeda
@Archomeda
Mar 18 2017 21:26
@lye still wasted? :smile:
ChieftainAlex
@ChieftainAlex
Mar 18 2017 21:56
:beers: :shipit: .. what could possibly go wrong
windwarrior
@windwarrior
Mar 18 2017 22:20
insert ballmer peak image
also, its weekend, let him also enjoy it :>
🤢
darthmaim
@darthmaim
Mar 18 2017 22:44
<3
turned it back on; not entirely convinced it won't start falling over again though
Archomeda
@Archomeda
Mar 18 2017 22:57
do you know what happened?
there was a cross-dc bandwidth issue, incoming requests piled up on the JS contexts which eventually led to V8 killing the process with an OOM error
I need to add a hard upper bound to the number of requests that can be waiting for a given cache key to resolve I think.
Michael Dougall
@madou
Mar 18 2017 23:00
:)
darthmaim
@darthmaim
Mar 18 2017 23:01
{
  "text": "unknown error"
}
on /v2/gliders
Darqam
@Darqam
Mar 18 2017 23:01
rip
darthmaim
@darthmaim
Mar 18 2017 23:02
<html>
<body>Error 42 (0:9001:4439): TacoProxy, transaction timeout {#267276}, {P+/Taco/Request+STS/1.0}, {10.51.69.118:56174 0 0 0 0}
<p>TacoProxy/302.6031806 Instance/4.501123093</p>
</body>
</html>
thats a new one
Michael Dougall
@madou
Mar 18 2017 23:03
thats awesome
darthmaim
@darthmaim
Mar 18 2017 23:05

What is it then? Something developed in-house?

Yes, it’s a simple reverse proxy that knows about our server infrastructure so it requires almost zero configuration.
It’s pretty great.

(source)
Archomeda
@Archomeda
Mar 18 2017 23:07
yeah... the guild log checker of my discord bot got a 502 as response
darthmaim
@darthmaim
Mar 18 2017 23:08
a bunch of endpoints return unknown error
Michael Dougall
@madou
Mar 18 2017 23:09
pls healthcheck api :-)
Werdes
@werdes
Mar 18 2017 23:09
i'm jiggling between 502, 400 or just timeouts :fire:
turning them abck off
thought they'd at least stay alive long enough for me to take a shower
rip
darthmaim
@darthmaim
Mar 18 2017 23:11
:(
Werdes
@werdes
Mar 18 2017 23:11
:sob:
might need to add rate limiting fml
Archomeda
@Archomeda
Mar 18 2017 23:33
that's /v2/items?
or /v2/wvw?
think it's items
Eearslya Sleiarion
@Eearslya
Mar 18 2017 23:33
Looks like items, the darker blue just below it is wvw
Archomeda
@Archomeda
Mar 18 2017 23:33
wow
darthmaim
@darthmaim
Mar 18 2017 23:35
there can't be that many sites crawling items?
beats me; we don't have detailed request logs for whatever reason
could rate-limit by IP and find out I guess
Darqam
@Darqam
Mar 18 2017 23:36
eesh that's quite a bit
Michael Dougall
@madou
Mar 18 2017 23:36
ip would be better
darthmaim
@darthmaim
Mar 18 2017 23:36
doubt that those are client side user requests, because they start as soon as you turned the api back on, probably someone who queued a bunch of api updates that all started at the same time now
maybe
Archomeda
@Archomeda
Mar 18 2017 23:37
^ i agree on that
probably some bad coding
error? let's queue that
better: queue it twice just to make sure
Archomeda
@Archomeda
Mar 18 2017 23:37
and in the mean time, fire a new request 1 minute later
error as well? let's queue that too
now we've two queued requests
etc
windwarrior
@windwarrior
Mar 18 2017 23:38
it would have made more sense if it was a commerce endpoint that got hammered
yeah because like, doesn't it only take like 700 requests to fetch all item data for a single language?
darthmaim
@darthmaim
Mar 18 2017 23:38
I think its just 250 or so
Darqam
@Darqam
Mar 18 2017 23:38
I just query once every 6 hours regardless of errors or not, so I doubt it's me...
yeah but like
Archomeda
@Archomeda
Mar 18 2017 23:38
i don't even hit /v2/items :D
windwarrior
@windwarrior
Mar 18 2017 23:39
materialheatmap hits it
but yeah, thats from JS
yeah but that makes like two requests too lol
Archomeda
@Archomeda
Mar 18 2017 23:39
that's client right?
10 requests/second sound reasonable?
darthmaim
@darthmaim
Mar 18 2017 23:39
I doubt there are 15k users sitting around hitting refresh while the api is offline
windwarrior
@windwarrior
Mar 18 2017 23:40
http://windwarrior.github.io/GW2MaterialHeatmap/ <-- that amazing error logging
Archomeda
@Archomeda
Mar 18 2017 23:40
i guess
windwarrior
@windwarrior
Mar 18 2017 23:40
Error: error
what happens if you get rate limited
does it just take a while
Archomeda
@Archomeda
Mar 18 2017 23:41
i'd say throw an error
windwarrior
@windwarrior
Mar 18 2017 23:41
or do you get like a response saying that you got rate limited
darthmaim
@darthmaim
Mar 18 2017 23:41
can we do 600/minute instead of 10/second, or how does it work?
i need to read the docs for the implementation we have
darthmaim
@darthmaim
Mar 18 2017 23:42
I guess you can just start with a low limit for now and we will see how it goes
i think it's burstable (e.g. 600/minute)
Archomeda
@Archomeda
Mar 18 2017 23:42
github uses 403 for rate-limited requests https://developer.github.com/v3/#rate-limiting
i was gonna send HTTP 402 Gem Purchase Required
darthmaim
@darthmaim
Mar 18 2017 23:43
perfect
Archomeda
@Archomeda
Mar 18 2017 23:43
lmao
Michael Dougall
@madou
Mar 18 2017 23:43
niiice
Darqam
@Darqam
Mar 18 2017 23:43
lol
ah dang there's HTTP 429 https://tools.ietf.org/html/rfc6585
darthmaim
@darthmaim
Mar 18 2017 23:44
isnt there a Payment required status or so?
Archomeda
@Archomeda
Mar 18 2017 23:44
ah right, i thought there was something
that's 402
i wonder why github uses 403 then, when 429 exists
but i've seen 429 somewhere else, but can't remember where exactly :(
Google Developers API uses this status if a particular developer has exceeded the daily limit on requests.
lol
apparently google uses 402 :P
Archomeda
@Archomeda
Mar 18 2017 23:59
something isn't right there
why did some requests drop and did /v2/items spike? D: