Failing Jira API requests made via OAuth2

We noticed that since mid-December some requests from one of our external services that talks to the Jira REST API via OAuth2 (3LO) are failing every so often. It does not seem to be a whole lot but it is enough to be noticed by customers too. If I look at our log histogram it is quite obvious. The issue seems to have started around Dec 9, getting less towards the holiday season and now ramps up again.

All the requests shown in the histogram fail with a 502 error similar to this one:

	"request": {
		"method": "POST",
		"path": "/ex/jira/<cloud-id-redacted>/rest/api/2/search"
	"response": {
		"status": 502,
		"statusText": "Bad Gateway",
		"data": "",
		"header": {
			"x-failure-category": "FAILURE_ORIGIN_CONNECTIVITY",
			"connection": "close",
			"x-trace-id": "af81906a525ad410",
			"x-xss-protection": "1; mode=block",
			"x-request-id": "af81906a525ad410",
			"transfer-encoding": "chunked",
			"expect-ct": "report-uri=\"\", enforce, max-age=86400",
			"x-frame-options": "SameOrigin",
			"x-content-type-options": "nosniff",
			"strict-transport-security": "max-age=315360000; includeSubDomains; preload",
			"date": "Tue, 05 Jan 2021 18:01:09 GMT"
	"timestamp": "2021-01-05T18:01:10.046999931Z"

Unfortunately, I do not think there is anything we could do on our end to fix or troubleshoot this.

Does anyone at Atlassian have some insights into this? Happy to share more details if anything else is required to troubleshoot.


1 Like

Hi @tbinna,

Thanks for reporting this. I have created


1 Like

Thank you, Dugald!

Just following up on this post because the issue on ACJIRA does not seem to get much attention.

I have just received a reply from the customer, which in turn got the following response from Atlassian regarding this issue:

Atlassian confirmed that this problem is caused by an application that registered the jiraEntityProperty module without keyExtractions element on our instance which affects multiple applications including Intercom for jira app. They implemented a permanent fix last week to resolve this and webhooks weren’t delivered from the period of December to January.

This leaves me with a couple of questions:

  • Is it possible to get some sort of report from Atlassian on the impact of this issue?
  • Can you tell how many other customers/instances were potentially affected by this issue?
  • How come webhooks were affected too?

We had and still have some reports from a few customers regarding some strange entity property behavior that we struggle to pinpoint. This makes me wonder if these customers may be affected by the same issue.