Slow JIRA API

[Alex Trandafir]

Hello,

Currently I am using JIRA with Agile( https://www.atlassian.com/software/jira/agile) in order to change a status of a ticket(ex: new, in review, in work, almost done, done, closed, released).

So I created a separate script(php) for parsing a ticket history, that adds to my counter just the time spent for a specific status(ex:in review + in work).

So far I used the API to extract all tickets(+ history) for a specific project from Agile, but, in time, the extraction process costs me a lot of time(for a project with 800 tickets - it takes almost 30 minutes to download all data)

Is there a better way to extract this data faster?
Thank you!

[Peter T]

Hi Alex,

even if you make it twice faster still as the number of tickets increases the time needed to get ALL tickets will go up. I suggest you don't get ALL the tickets but narrow them down to "in review" and "in work". The REST API supports JQL: https://developer.atlassian.com/display/JIRADEV/JIRA+REST+API+Example+-+Query+issues

[Alex Trandafir]

Than you Peter,

But I really need all data downloaded in a separate DB in order to create different reports.
I was just wondering if there is a better way to do this; my approach is:

1. first, I run a curl request in order to get all issues for a specific project:

curl -u 'user:pass' -X POST -H 'Content-Type: application/json' --data '{"jql":"project = '.$project_id.'","startAt":0,"maxResults":1,"fields":["key"]}' 127.0.0.1/search

2. Then, I parse them in batches of 50. For each, I create a new curl request:

curl -u 'user:pass' -X POST -H 'Content-Type: application/json' --data '{"jql":"project = '.$project_id.'","startAt":'.$offset.',"maxResults":'.$step.',"fields":["key", "summary", "issuetype"]}' 127.0.0.1/search";

3. Inside of previous for-each, i create again a new curl request in order to get the issue changelog history

curl -u 'user:pass' -X GET -H 'Content-Type: application/json' 127.0.0.1/issue/{$issue->key}?expand=changelog

[Boris Georgiev [Appfire]]

I would suggest upgrading the logic in your script to get only issues that were updated and created since the last time the script was run by including "updated" field in your JQL (i.e. updated >= 2014-02-04 AND updated <= 2014-02-05), which will narrow down the results. You just have to remember the date/time the script was run.

I think no solution no matter how fast it is will scale when the issue number increases over time.

[Alex Trandafir]

Thank you Boris. I think this should make it faster. For the moment I need only the tickets from last month, so, maybe 50 max per script run.
I really don't know why I am not using already the "updatated" option - perhaps not knowing all api possibilities.

Thanks again.
I will be back with more info as soon as I update the script.