any tips on debugging slow queries- I’m seeing 2-3...
# prisma-whats-neww
wesbos
02/09/2018, 4:41 PMany tips on debugging slow queries- I’m seeing 2-3 seconds on some queries
h
huv1k
02/09/2018, 5:05 PMyou can track queries
w
wesbos
02/09/2018, 5:05 PMlike in playground?
wesbos
02/09/2018, 5:05 PMIt just tells me the entire query is slow, which I know.. I want to know why it’s slow
http://wes.io/pQhG
h
huv1k
02/09/2018, 5:06 PMyea, but you need to have yoga for example
huv1k
02/09/2018, 5:06 PMand how looks your resolver?
w
wesbos
02/09/2018, 5:07 PMwesbos
02/09/2018, 5:07 PMit’s jsut two queries..
h
huv1k
02/09/2018, 5:09 PMdo you use prisma cloud or local?
w
wesbos
02/09/2018, 5:11 PMcloud
n
nilan
02/09/2018, 5:15 PMwhere are your server and prisma cluster deployed?
what's your
prisma-bindingnilan
02/09/2018, 5:15 PMnote that the public cluster employs a rate limit of 1 request/s
w
wesbos
02/09/2018, 5:16 PM@nilan “prisma-binding”: “1.5.7"
wesbos
02/09/2018, 5:16 PMus1.prisma.sh
wesbos
02/09/2018, 5:17 PM1 request per second eh?
wesbos
02/09/2018, 5:17 PMso if I want to query something, and then update it, it will take at least 1.1 seconds?
wesbos
02/09/2018, 5:18 PMor if 60 people visit my site, it will take 1 min for the 60th person to view the site?
n
nilan
02/09/2018, 5:36 PMI collected more information about the rate limiting here: https://github.com/graphcool/prisma/issues/1863
nilan
02/09/2018, 5:37 PMcan you share your resolver?
nilan
02/09/2018, 5:37 PMah see it now
nilan
02/09/2018, 5:40 PMhow do you deploy your GraphQL Server? there might be high latency between your server and the Prisma cluster
w
wesbos
02/09/2018, 5:41 PM@nilan the yoga server? its local
n
nilan
02/09/2018, 5:45 PMgotcha! could you try
ping us1.prisma.shw
wesbos
02/09/2018, 5:45 PMPING cluster-public-us-west-2-734360691.us-west-2.elb.amazonaws.com (52.89.189.152): 56 data bytes
64 bytes from 52.89.189.152: icmp_seq=0 ttl=235 time=81.243 ms
64 bytes from 52.89.189.152: icmp_seq=1 ttl=235 time=81.571 ms
64 bytes from 52.89.189.152: icmp_seq=2 ttl=235 time=82.323 ms
wesbos
02/09/2018, 5:46 PMits not all queries, just that one
wesbos
02/09/2018, 5:46 PMso if you are rate limiting it- that would explain it
n
nilan
02/09/2018, 5:48 PMright!
w
wesbos
02/09/2018, 5:48 PMthat is pretty agressive- is the only way around that to host locally?
wesbos
02/09/2018, 5:48 PMi can’t really see me doing a video series with such a laggy UI
h
huv1k
02/09/2018, 5:50 PMuse local prisma instance?
w
wesbos
02/09/2018, 5:50 PMI just asked that
wesbos
02/09/2018, 5:51 PMthe whole docker thing is a nightmare in tutorials
wesbos
02/09/2018, 5:51 PMseems easy in concept but people get so frustrated and ask for refunds because they can’t get past the local installs
d
divyendu
02/09/2018, 6:06 PM@wesbos:
Hi 👋,
1. You should be able to set your own cluster on DO and use it as a shared cluster for videos i.e. no rate-limiting, you managing the infra... in theory this should be possible with
prisma add clusterl
lawjolla
02/09/2018, 6:22 PMI have the same issues as Wes, but figured it was part of the public cluster. I'll be buying your course, and I wouldn't mind a sleight of hand deploying locally for the video but teaching the cloud... or it could be viewed as an opportunity for sophisticated optimistic UI updates! 🙂
n
nilan
02/09/2018, 6:30 PMWe'll look into lifting this limit over time, but we want to make extra sure that the development cluster is protected from malicious/accidental request bombarding 🙂 Thanks for your feedback on this one, everyone 🙌
b
Ben
02/09/2018, 7:46 PM@wesbos what do you think about a limit of 60 requests/minute. I think that would feel less laggy. https://github.com/graphcool/prisma/issues/1863#issuecomment-364542184
w
wesbos
02/09/2018, 9:04 PM@divyendu thank you - I did’t think of hosting my own cluster. I could obviously have many apps on that?
👍 1
n
nilan
02/09/2018, 9:06 PMyes 🙂
m
max
02/10/2018, 10:19 AM@wesbos let me know what you end up doing 🙂 I'm also seeing slow response times between server and prisma (co-located in same AWS region). If I run tons of queries off a data heavy page it slows to 10 seconds per request. I hope it's just the throttling.
3 Views