JIRA rounds 'updated' date differently across different REST APIs

Perry Krug

Jira v8.5.7

 

It appears that the date/time returned from a JQL search to the REST API differs from the date/time returned from the issue directly:

 

Using the REST API to search:
<a href="https://issues.couchbase.com/rest/api/latest/search?jql=key+in+(%22CBSE-8965%22)+order+by+updated+desc&fields=updated&startAt=0&maxResults=2000" target="_blank" rel="noopener noreferrer nofollow">https://issues.couchbase.com/rest/api/latest/search?jql=key+in+(%22CBSE-8965%22)+order+by+updated+desc&fields=updated&startAt=0&maxResults=2000</a>
The result is:

{
expand: "names,schema",
startAt: 0,
maxResults: 1000,
total: 1,
issues: [
{
expand: "operations,versionedRepresentations,editmeta,changelog,renderedFields",
id: "139745",
self: "https://issues.couchbase.com/rest/api/latest/issue/139745",
key: "CBSE-8965",
fields: {
updated: "2020-10-08T16:42:27.000-0500"
}
}
]
}

See that the 'updated' field is "2020-10-08T16:42:27.000-0500"However, if I use the REST API to fetch the full document contents:
<a href="https://issues.couchbase.com/rest/api/latest/issue/CBSE-8965" target="_blank" rel="noopener noreferrer nofollow">https://issues.couchbase.com/rest/api/latest/issue/CBSE-8965<br><br></a>Then the return 'updated' field is: updated:&nbsp;"2020-10-08T16:42:27.259-0500"<br><br>So it would seem that JIRA is rounding the date in different places

Andy Heinzer

Hi Perry,

Thanks for pointing out this problem.  I can see how this difference could cause confusion.  I found that this has already been documented as a bug back in JRASERVER-28238.  I have updated this bug so that it is more in line with our standards.

Sorry to report that there does not appear to be a clear workaround for this.  However I suspect this has been prioritized low because Jira never displays date/times to this level of precision within the UI.  That said, I didn't expect the values to be different here between these two calls. 

When Jira is searching for issues, it has to rely upon the indexes created for these issues.  Jira uses Lucene to do this. I suspect that in order to optimize the search indexes, these millisecond values might be getting dropped from the actual value when the indexes are created.

I hope this helps.

Andy