How to deal with outages not acknowledged by Atlassian?

Question

We use Jira and Confluence hosted in the cloud.  I'm "regularly" running into confluence outages that can last minutes or hours.  First thing I do when Confluence is acting really slow or not loading at all is check the Atlassian network status page, which nearly almost always don't report any outages when I do.

I check in different browsers and different PC's and confirm that the Jira site loads and responds fast but the Confluence site doesn't load or has various errors.

I go to report the issue using the link "Report a problem" and instead of making it trivial to report outages so that their system can flag user reports, they instead determine I need to report this to my admin on a Friday evening and the admin has to turn around and report an outage. There's absolutely nothing that the admin needs to do I couldn't report. FFS, that's just horrible customer service when trying to report and outage.

Anyway, this may come off as a rant about how unreliable the cloud service is, but I'm really asking if there is another way to let them know about network issues that isn't going to take half a day or a weekend to report?  I know it will just sort itself out, but it's been 3 hours now.  I'm at my wits end with lost productivity with this happening for hours at a time. I'm more inclined to ask our admin when our renewal is (to move off of it) than to have him file a support ticket about poor reliability that just ends up with US doing the homework and logging all the times its been garbage.

Michael Rex_cwzb8i · Answer

Stephen,

Clearly, I didn't make myself very clear. Atlassian is dropping the ball on analytics. There is no need for an admin to report a complete outage. It's easily done in a few minutes by checking from other PC's in other locations. It just needs to be a simple reporting, no elaborate details required. No tickets, no back and forth. This is a numbers reporting issue. There's a difference between opening a support case and literally what it says, "Report a problem".

With analytics, they'll have a very low amount of reports per hour and then when there is a major issue, there will be spikes in user reported outage reports. Boom, a threshold is reached and then the OPS team is notified of a real world customer facing issue, WITHOUT TAKING HOURS, WITHOUT INVOLVING ADMIN FOR TRIVIAL THINGS.

I already had cases opened with support for poor performance and being added to bug tracking (no progress was ever made and it looked like they just wanted to look like progress was made when they just kept saying to use workarounds).

Go see what a site like downdetector.com looks like in terms of user reporting. https://downdetector.com/status/confluence/

Anyways, my post was out of frustration that this happens for hours at a time, for several times a year. No other cloud service we use has this many outages. I wouldn't have a problem if it took 15 minutes for them to confirm an issue, but its hours, multiple times now. That's frustrating, especially because they can and need to do better.

Bill Bailey · Answer

This is a fundamental aspect of Cloud apps -- you are at the mercy of the companies server and the CDN it is running over.

I also think that their reporting model was just carried over from the server days where admin could address a lot of issues, and not rethought for the Cloud era.

Just add this to the long list of reasons I do not recommend the Cloud for enterprise solutions.

Ste Wright · Answer

Hi @Michael Rex

The reason https://support.atlassian.com/ requests an Admin to report the issue, is because it might be something else beyond an Atlassian issue - such as...

* A Space with a large amount of attachments
* A Page with significant complexity - such as large amount of JQL filters
* ...etc

Your Admin might want to do some of this research first, prior to reporting the issue to Atlassian.

As you say, there is an active Confluence incident on https://confluence.status.atlassian.com/

You do have to remember there are hundreds of thousands of instances globally across a range of data centres - so rectification might take a little while at times!

Ste