These are chat archives for reactioncommerce/reaction

17th
Apr 2018
MWPAWAR
@MWPAWAR
Apr 17 2018 13:31
Hi Everyone!
I anybody faced this issue before.
I added simple:rest meteor package to enable HTTP calls and created some methods in the custom plugin.
The API hook is working correctly. Everytime I call the API, it invokes the correct method. I was trying to createUser in the same method. But currently, I am facing following issue:
Reaction: Meteor.userId can only be invoked in method calls or publications.
Does anybody have any idea?
Loan Laux
@loanlaux
Apr 17 2018 13:35
Looks like you'll have to create a Meteor method that wraps createUser @pawar-mahesh
MWPAWAR
@MWPAWAR
Apr 17 2018 14:09
This worked. Thanks @loanlaux .
Loan Laux
@loanlaux
Apr 17 2018 14:09
You're welcome!
MWPAWAR
@MWPAWAR
Apr 17 2018 14:09
But just a curious though why it not worked previously.
Loan Laux
@loanlaux
Apr 17 2018 14:09
That's a Meteor limitation, @pawar-mahesh
MWPAWAR
@MWPAWAR
Apr 17 2018 14:12
:+1:
Daniel Honig
@dhonig
Apr 17 2018 21:38
So is there an easy way to disable migrations?
ragboy
@ragboy
Apr 17 2018 21:59
We are having an issue with site in production after an import, the instances are running something and hogging cpu, slowing down site to a halt. Is there a meteor TOP or some way to view the processes running to trace what they are doing?
Daniel Honig
@dhonig
Apr 17 2018 22:29
@ragboy what is your deployment like?
Docker based?
ragboy
@ragboy
Apr 17 2018 23:42
yes, we got through it by scaling up mongo much bigger than we expected we needed.
Daniel Honig
@dhonig
Apr 17 2018 23:43
Whats much bigger and in a replicaset of course?
ragboy
@ragboy
Apr 17 2018 23:43
But the import was done, and then many jobs for dealing with the images and stuff, and they keep loading async without any max.
so we normally use M30 without issue.
We have ECS scaling containers at 30+% cpu, and the underlying instances at 35%.
we maxed out at 100 containers and 18 instances, with over 500 connections and had to go to M100 mongo instance to keep up.
otherwise the site just halted, and there was no traffic, maybe 20 users, it was all from the jobs on the import.
Daniel Honig
@dhonig
Apr 17 2018 23:45
nice
in a similar situation now :)
ragboy
@ragboy
Apr 17 2018 23:46
after several hours of this, its finally starting to settle down.
but this is seriously inefficient.
Daniel Honig
@dhonig
Apr 17 2018 23:46
Using a rest endpoint to receive products?
I'm getting about throughput of one product a second on a rest endpoint
ragboy
@ragboy
Apr 17 2018 23:51
the import was done, that did hammer the instances hard, but it was the the job processing of the images is what killed the system
image.png
you can clearly see where the import started. ;-)
yes, we are using atlas and a rep set, so we can scale veritically on demand.
sorry, missed that question.
ragboy
@ragboy
Apr 17 2018 23:58
I manage lots of different systems in AWS, mage 1 and 2, custom PHP, and then 2 reaction systems, this is the first major scale event I have dealt with. Def an area of concern. ;-)