Atlassian connect rate of requests limit

[Daniel Turczanski]

Hi,

Our plugin makes requests to Atlassian Connect API and adds metadata to Entity Properties in all the issues on cloud instances.

Recently some of our clients have reported that the rate we hit their cloud instances causes slowdowns of JIRA.

What is the rate limit we should adhere to?

This is an urgent question we need to address.

Thanks,

Daniel

[Robert Massaioli _Atlassian_ · Accepted solution]

Forty fields do sound like a little bit of data to me. But reacting to webhooks by posting data back to JIRA should be acceptable. Maybe, if there are problems posting back to JIRA, then you should create a message queue that sends back the webhook responses 1-5 at a time.

I suspect that you might be encountering issues due to bulk updates. In a bulk update, you are going to get hundreds of post messages in a short amount of time. If you respond to JIRA immediately you might overload it.

How many issue property updates per second is that averaging to before you notice problems?

[Daniel Turczanski]

Limiting to 5 connections fixed a lot of timeouts, we still get some but they are really infrequent. Is there any plan for improving the performance of JIRA. I'm pretty sure it used to be better i.e. we started getting the timeouts over a month ago.

[Daniel Turczanski]

1. it's mainly /rest/api/2/issue/ISSUE_ID/properties/ADDON_KEY
2. we are hitting it each time the change is made in JIRA issue based on webhooks we receive
3. It's a json request, not much data there, maybe 40 fields

[Robert Massaioli _Atlassian_]

Please keep in mind that any answer that we give you now will become invalid within the next year: Atlassian is always changing our backend systems to make them more performant.

That said, if your add-on alone is hitting over 5 requests per second then I suspect that you will begin to see a twinge of a slowdown. If you start hitting 10 requests per second then I suspect that the slowdown will be quite noticeable. (Depending on which instance you hit and what rest resources you are hitting and with how much data)

Can you answer these questions:

1. Which JIRA REST resources are you hitting?
2. How fast are you hitting these rest resources accross the entire customers instance (requests / second)
3. How much data (approximately) are you sending in each request and what is the structure of that data?

Cheers.

[Jon Bevan [Adaptavist]]

@Robert Massaioli _Atlassian_ can you give us an update on those rate limits given that they are now about 1yr out of date?

Additionally, is there a recommended limit on the number of concurrent requests we should be making to any given Jira Cloud instance?