Moved to Elasticsearch - Change in function of Activity and Content Feeds?

[Emediate]

After so many years of putting it off I've finally migrated to Elasticsearch. Can't believe I waited this long, as choosing a local AWS server has seriously sped up general use of the forums and functions.

However, since I did this some members are complaining about a change in the way Unread Content, and View New Posts is working.

Apparently, prior, these would just show the last post in a thread, whereas since enabling Elasticsearch, Unread Content will return every single unread post in a thread.

I did at the same time tweak the settings of the Search functionality so I don't know if it was my doing, or what the default values should have been.

My settings:

Not sure if it was my doing, or just a change in how things work once Elasticsearch is enabled. Any guidance appreciated! 🙂

[Marc Stridgen]

It sounds like the stream was changed. You need to edit the stream and make sure you have it set to "content only"

[Emediate]

Thanks @Marc Stridgen.

These are the current settings:

I can't see which I should change?

Here's another example from a member:

Unread Content 

 

You'll notice it's set to Topic under Content Types, yet you can clearly see two posts in the Vinyl - Currently Spinning thread showing only minutes apart.

[Marc Stridgen]

In the first instance, please rebuild the search index from system>settings>search using the button provided in the top right. Once this is done, please let us know if you see the same problem

[Emediate]

26 minutes ago, Marc Stridgen said:

In the first instance, please rebuild the search index from system>settings>search using the button provided in the top right. Once this is done, please let us know if you see the same problem

It literally just finished 1-2 hours ago.

[Marc Stridgen]

40 minutes ago, Emediate said:

It literally just finished 1-2 hours ago.

Before I look further, could you confirm that wasnt the initial rebuild, and was an attept after the initial rebuild?

[Emediate]

We did one build, when moving on to the Elasticsearch Server. Then another yesterday that finished earlier today when we changed the Elasticsearch Server location. I've triggered a new one now. That will take 24 hours.

[Marc Stridgen]

No problem. Please let us know as soon as this is complete if you still have the same problem. We can then escalate if needed

[Emediate]

Just checking back in. Search index rebuilt, and exactly the same behaviour as described.

[Marc Stridgen]

I have created a ticket for you on this, and someone will be in contact as soon as possible

[Emediate]

Any chance this can be picked up again soon please?

My members are losing it on this issue 🙂

[Marc Stridgen]

This is in our developers queue at the moment. Support of an advanced nature can take some time to address

[Emediate]

I appreciate that, I really do. Can we enable 2FA and have the developers activate it when they log in?

It leaves a site vulnerable for an extended unknown amount of time that's all.

[Randy Calvert]

38 minutes ago, Emediate said:

I appreciate that, I really do. Can we enable 2FA and have the developers activate it when they log in?

It leaves a site vulnerable for an extended unknown amount of time that's all.

I put the account used by IPS into a new group called “Admin Backup”. That account uses security questions for login. I stored the answers for each in my client area notes. 

It lets me keep 2FA on for normal admins and IPS can still access with the security questions. 

[Jim M]

1 hour ago, Emediate said:

I appreciate that, I really do. Can we enable 2FA and have the developers activate it when they log in?

It leaves a site vulnerable for an extended unknown amount of time that's all.

This would need to stay disabled while our developers look into this or it will cause extreme delays in us looking into your issue, I'm afraid.

[Emediate]

1 hour ago, Randy Calvert said:

I put the account used by IPS into a new group called “Admin Backup”. That account uses security questions for login. I stored the answers for each in my client area notes. 

It lets me keep 2FA on for normal admins and IPS can still access with the security questions. 

@Jim M would the above be suitable for escalated support cases?

[Jim M]

23 minutes ago, Emediate said:

@Jim M would the above be suitable for escalated support cases?

Yes, if you provide the security answers in the notes section of the Client Area Access Details

[Randy Calvert]

If you do this, 100 percent confirm you login as the support account using the info on file including the security questions. Make sure the info is spot on to avoid delays in getting help. 

[Matt]

Just as a note to future readers of this topic, the issue looks to be by the Elastisearch provider not allowing access to the task and update by query APIs.

If you get any odd issues with multiple entries where there shouldn't be in streams, it's almost always update by query not working or being blocked by the host.

[Emediate]

Thanks for the update. In that case, Bonsai is not suitable for an Elasticsearch/Invision integration as you've discovered (on anything less than their $700/month plans).

Going direct with Elasticsearch Cloud, they don't seem to have a version that is supported by Invision:

What is Invision's advice here when it comes to integrating with Elasticsearch please?

[teraßyte]

Version 7.17.5 in the dropdown should work. It's higher than 7.2.0 and lower than 8.0.0. 🤨

Edited August 16, 2022 by teraßyte

[Emediate]

Maths was never my strong point, but unless I'm missing something blindingly obvious, 7.17.5 is not higher than 7.2.0?

[Jim M]

7 minutes ago, Emediate said:

Maths was never my strong point, but unless I'm missing something blindingly obvious, 7.17.5 is not higher than 7.2.0?

17 is greater than 2 🙂 

[Emediate]

LOL - I even consulted my wife for an opinion on this one.

We both feel stupid now.... 🙂 

Thanks for the clarity.

[Emediate]

For future reference, with this information provided by Invision, I have implemented a local installation (not Bonsai) of Elasticsearch and can confirm the original issue is now resolved. Thanks @Matt/ @Marc Stridgen.