Why is cloud confluence indexing disabled?

[Rob Woodgate · Accepted solution]

Hi Shane,

It's one of the restricted functions in Cloud, presumably to prevent lots of high-load events being kicked off at the same time and affecting performance.  You can ask Atlassian to re-index it for you.

[Shane Clementson]

Thanks for the response Rob.  I presumed the same thing, but found it interesting that there does not seem to be any explicit statement confirming this is the reason - none that I could find anyway.  In any case, if this is the reason, the current advice of having to submit a request to re-index content seems a little primitive given the search function is an essential part of almost every app, and especially important for a wiki.

I guess a solution I would prefer to see on Atlasssian’s “Restricted Functions in Confluence Cloud” page is something like, “Content indexing for Confluence Cloud is a System Administration function not available to Cloud users.  Instead, Atlassian will re-index content for a Confluence instance each time it is found to increase by 5% (or a minimum once per month).  The scheduling of this indexing function will be entirely at Atlassian’s discretion to ensure that there is no impact on system performance.”

The 5% and min 1 month frequency, are just notional thresholds for this example (I expect Atlassian would have the data to decide what thresholds would be best to use on the searchable content for an instance), but I’d have thought something along these lines would provide a reasonable balance that facilitates a generally useful search function and preserves performance.  i.e. Users in most cases should be able to either: find the most recently edited pages in the “All Updates” list (i.e. containing content being actively worked on but not indexed yet), or otherwise search for (and find!) older content without having to first manually submit a request  to Atlassian.

[Rob Woodgate]

Hi Shane,

By all accounts this functionality was present in earlier versions of cloud.  Vote for this ticket https://jira.atlassian.com/browse/CONF-33400 to help bring it back

[Shane Clementson]

Thanks Rob.  Looks like there's plenty of other people that are equally baffled as to why Atlassian removed it and why the company can't find a better indexing solution that provides reasonable search functionality.  I've marked the answer as accepted and put the question directly to Atlassian in my support request.

[Paul Abramson MD]

I am also baffled by this. We are having extremely degraded indexing performance for a very small wiki. Adding new pages & they aren't accessible via mobile app or searchable by anyone. Nor are they picked up by "list content by label" or similar functions.  Yet there are no incidents mentioned on the Atlassian cloud status page.... Boo.

[Shane Clementson]

Hi Paul,

I thought I'd paste in the response I received from Atlassian Support in case it's helpful for you:

"All of Confluence's searchable content is added and sent to the index when they are created. A reindex is when we ask Confluence to read through the entirety of the content from it's database (everything that is Searchable) and create new documents from this content which is sent to the index. Creating new documents includes running Extractors over each piece of content and 'content' includes Pages, Blogs, Comments, Attachments and more is an expensive process to run.

Disabling anyone besides sysadmin users are intended to protect the instance from crashing due to memory errors. Reindexing is only necessary when there are changes on how one or more fields analysed, requiring a new field, a full site or space level import where it will run after completion as part of the import process.

If you are running into a behaviour that you feel that a reindex can solve. Please let us know the issue and provide us with your site details."

[Paul Abramson MD]

My problem is that over the weekend new pages I create are NOT being indexed, aren't searchable in the regular search box, and don't show up in "show content by label" lists (but DO show up in child pages list). Of course all shows "online and functional" on the status page for servers...

[Rob Woodgate]

Hi Paul,

The status page provides up-to-date information about the general status of the Atlassian servers.  It's not intended to indicate the presence of an individual issue with one customer's instance. There are thousands upon thousands of cloud instances running at any one time, so it's not feasible to have a status page that indicates to every customer that there is a problem, just because one customer might have an indexing issue.

Atlassian will be happy to kick off a re-index process for you if you ask them, and in my experience they'll do it very quickly.

[Rob Woodgate]

Hi Shane,

Thank you, it's always good to have an official response available for anyone else who has this question.

[krugerf]

I have the issue today with new pages not being available, was fine yesterday. what is the root cause of the issue?