James Ponting mentions planned work of separating user indexing from a full index, which should allow a less disruptive workaround should it be required in the future.  Is there a separate issue that I can follow to track the progress of that separate user index?  

We experience the issue where new users are occasionally not available for an "@" mention.  Steps I need to follow to address are doing full site re-index and then going into People directory and forcing username to come up through the directory.  Steps are described in this support article: https://confluence.atlassian.com/confkb/unable-to-mention-certain-users-in-confluence-779165995.html.  The steps work, but I do not like having to do a full site re-index each time it occurs.  

We are in version 8.7.2, so do not have the 'UserIndexingUpgradeTask'  task option available to us.  

Hi 4ab64ba38276,

You've touched on all the reasons we've taken this direction on the issue.

The behaviour you are describing is not the same behaviour as was detailed in this issue, and demonstrates how this issue is acting as a catch-all. I've regularly taken the approach of requesting additional information directly on tickets, but this is not an issue where that is appropriate. There are a number of different behaviours reported here that need to be individually investigated.

Closing the issue directs customers back to our Support team who can properly investigate and, importantly, identify the actual issue being faced. They can then capture the information required for development investigate and fix it. I've also included additional debug logging in the hope of reducing the time to getting an answer when working with Support.

Whilst I appreciate that this isn't the answer that you wanted, I can't achieve the outcome you're chasing without taking this action.

Regards,
James Ponting
Engineering Manager - Confluence Data Center

This is ridiculous.  So now we have to wait for another ticket to be opened, voted on "gather interest", etc. so Atlassian can wait another 10 years to NOT solve a problem?

I cannot reproduce this as will, but when it happens it continues to happen (for that user).  If you simply would ask for help (like here!), we could provide more information.  I see no comment or added to ask for any information/logs/etc/.  We do not bother further reporting on the issues, because we have (had) an issue already open on this.  For us, a large user of the tools with over 7000 users, the "edge case" is not that much on the edge.  We get users reporting this problem weekly.  It is VERY annoying and has been with us for a VERY long time.  This is the kind of "support" that drives customers nuts.  You have a decent product, but don't expect to keep customers forever when this is how you treat your largest customers (as an "edge" problem, it affects your largest customers most often). 

Hi All,

I'm marking this ticket as Closed and wanted to provide an explanation as to why.

This issue was originally opened by a Confluence developer in 2013, reporting a regression of CONFSERVER-6404: Users from LDAP do not appear in people directory or search results until they have logged in (fixed in Confluence Server and Data Center 4.3.2, 5.0). In this original issue, any and all users who were added to Confluence by LDAP sync would not appear in the People Directory or mention list unless they had logged in. Upon investigation (in 2013) we determined that the issue was actually an unaddressed edge case rather than a regression. At some point since then, the reported regression/edge-case appears to have been fixed, but unfortunately this ticket has remained open.

Recently this issue was bumped to a higher priority due to the customer interest in the issue. As a result, my team have been investigating the issue, but unfortunately were unable to reproduce the behaviour as it was already corrected. Working with our friends in Support, we attempted to obtain additional information and replication steps. During this process, it became apparent that this issue has begun to serve as a "catch all" for various behaviours where a user is unable to be mentioned, or does not appear in the People Directory. This was compounded by there being this known issue that seemed to describe the behaviour, and as such no additional investigation was conducted. Over time many tickets were associated with this issue without additional information gathering.

As we don't have any actionable information, and there are multiple undefined issues linked, I'm closing this issue. I acknowledge that there remain issues with users being mentionable in Confluence, and would encourage you to contact Atlassian Support if you're able to reproduce these issues so we can investigate further.

Additionally, you can enable additional debug logging on the following classes when reproducing the issue to assist with investigating

1. com.atlassian.confluence.user.crowd.UserIndexingListener - The event that should index users being synchronised
2. com.atlassian.crowd.directory.DbCachingRemoteChangeOperations - The sender for the above event

We also have some planned work to seperate user indexing from a full index, which should allow a less disruptive workaround should it be required in the future.

Thanks,
James Ponting
Engineering Manager - Confluence Data Center

This happens even with folks who have logged in.  It is sporadic and seems to happen quite frequently.

We are facing this issue as well. Random users with all relevant permissions can't be found when trying to add to a comment or page.

I will request this to be higher prioritised for a fix. 

This is not just "a user has not logged in problem", but can happen to users that have logged in.  This seems to be a 'random' problem and thus the bigger the org, the more likely to run into this problem.  I know that we have users running into this problem close to daily.  So basically you are choosing to put to lower priority a problem affecting your largest customers the most.  This problem was first recorded over 9 years ago.  How about finally fixing it!?

Confluence 7.13.7

Problem still exists.

As @Jaedo Aum said: A fundamental problem resolution is required 

Seeing that this is on long term backlog, after updating to a months-old enterprise release (7.4.5) exposes our instance to a fundamental issue like this, is pathetic. 

Confluence 7.4.1

We have the same problem as @Jaedo Aum mentioned and agree on the increasing frequencies.