Elastic Search module

[vincentdenkspel]

Hi,

Lately there is activity on the forum from people who are trying to get the Elastic Search module working.
Most of them, if not all, did not succeed and are looking for alternatives.
I'm not a developer but from what i understand one of the problems is because the module is still in beta.
I think (most) alternatives will cost money.
I was thinking: instead of looking for paid alternatives this money could also be use to help funding so complete
the project so we would have a working version of the ES module.

I already supported the initial crowdfunding project, but i would certainly be interested to invest a little bit more.

So my questions are :
- Would other people be interested in doing so ?
- Would one of the main developers be interested to complete the module ?
- If so, can a indication be given how much it would cost to complete the module ?

Looking forward to hear what others think of the idea ?

Best regards,

Vincent

 

Edited June 17, 2019 by vincentdenkspel

[datakick]

I'm not sure if this module is worth the trouble. I mean, it can be extremely useful for anyone with huge amount of products, but it's not for an ordinary merchants. I believe there is less than 5% of the tb community that actually have needs for it. And these merchants can usually afford to purchase some commercial alternative, one that comes with support. Or they can hire developer to set this module for them.

I'm not against another crowdfunding project, but it would be better if it was something useful for entire community. Maybe some advanced order editing module, dran'n'drop cms page builder, shortcode system to use in html editor, or whatnot.

[vincentdenkspel]

2 hours ago, datakick said:

but it's not for an ordinary merchants.

I don't now if this is true. If i'm not mistaken one of the functionalities of the module was the 'advaced search' functionality ( product x in category y with features K AND L AND M AND N).
For me this was the main reason to fund the project with a lot of money.
I think this is a feature that is not only interesting for large shops but also for smaller ones.

Sure, you can buy a module for that (from PrestoChangeo ($250) and PrestaModule (€280), Elastic Search module € 200)) but buy these if you could have a TB developed one.
For me that reason the buy a gig.

Although i bought a large gig with extra development time included (so my money is not waisted) i can imagine that other people who invested less money will not be very happy.

That said i hope that TB will does other crowdfunding projects in the (near) future, because i really think it can be beneficial for TB as a project, for the developers and for the community as a whole

[dynambee]

When mdekker created the ES module a lot of time was spent to make it better than the modules that were already out there for PS.

Each of the PS modules that existed at that time did some things well and other things poorly. The goal was to make the TB ES module the best of all worlds, one that did each aspect of ES functionality in the best, fastest way possible. It took a lot of time, effort, and money to get the project to the point that it is at now. It would be a real shame to abandon such an advanced, high performance module at this point.

As far as the users who have been having trouble with it, there seem to be two. One is me, who has been testing and troubleshooting various aspects of the module, and the other doesn't seem to be able to get his ES server installs to work. I suspect he is using incompatible versions of the ES server or that something is misconfigured with his ES installation or the server that ES is running on. He doesn't really reply to my messages asking him to use compatible versions of Elasticsearch Server so it is quite difficult to help him.

As far as if the module is worth the trouble or not, I would say that it absolutely is. Once it is operating correctly it would just be a case of installing the module and turning it on. For users of something like Cloudways they don't even need to install Elasticsearch itself as it already exists on Cloudways servers. The module is very easy to install and configure and the resulting search performance is staggering.

[wakabayashi]

In my opinion ES was a too big project for this small core team. There is no doubt, ES offers great stuff. But as we can see, the set up is quite difficult. Michael is no more on the team and nobody has the time to look at it. I agree with @datakick that ES is mainly for big merchants. The inbuilt search works for smaller quite ok IMO.

• We still haven't a clean checkout in the core.
• Order Editing in the BO is a pain for many merchants. 
• Adding custom content to a page is too hard for most merchants (shortcodes needed)
• Controll of emails (what is send and how does it look like) is very hard.

In my opinion this points are much more important for now. Imagine how much time a coder needs to understand the ES module now. As far as I remember there is no support for facetted navigation at all. Is this correct @dynambee?

@vincentdenkspel The function of searching products with feature X, Y, Z - what is the difference to a filter?

[vincentdenkspel]

@wakabayashi There is no difference. However as far as i know there are two module that can do AND filters (meet all criteria). The rest do OR filtering.
This is not a problem as long as as you do select a option for each criteria. However, i need multiple selections. 
The modules which can do that costs between €250 and € 300 with one your free upgrades.

Why pay that if it could be done with the ES module

 

[wakabayashi]

@vincentdenkspel Are you talking about "Advanced Search 4"? I am using this one for filtering. In general I get your point. 

I wonder how many people are using ES now? I believe we would need a lot of investments to make this project really succesfull. Not only the coding part, but also the documentation & support... I just don't see the guy, who will do that. Cause as far as I can see ES is not something any random developer can do. You really need a quality coder for that. And as I stated above, IMO its better, if the best coder focus on topics, which are important for the majority of users.

Have you looked at services like doofinder or algolia? For example doofinder offers a free ps module and 30 days of free testing. 

[dynambee]

38 minutes ago, wakabayashi said:

In my opinion this points are much more important for now. Imagine how much time a coder needs to understand the ES module now. As far as I remember there is no support for facetted navigation at all. Is this correct @dynambee?

No, one of the good things about the ES module was that it supported proper faceted navigation, and in a way that worked extremely quickly.

I should have a better test site available for the ES module soon, one with a few thousand products on it. As far as I know the ES module worked fine with early versions of TB, there was much testing done. Later versions of TB have made changes that have caused incompatibilities with the module.
 

38 minutes ago, wakabayashi said:

In my opinion ES was a too big project for this small core team.

Part of the agreement when the module was funded & created was that it would be supported by the TB team along with other TB-created modules. It seems very unreasonable to renege on this the very first time problems are reported.

[wakabayashi]

2 minutes ago, dynambee said:

I should have a better test site available for the ES module soon, one with a few thousand products on it. As far as I know the ES module worked fine with early versions of TB, there was much testing done. Later versions of TB have made changes that have caused incompatibilities with the module.

Yeah that would be amazing. You seem to be the guy, who knows this module the best. 

Of course my view on the situation would be different if michael would still be on the team.

[datakick]

3 hours ago, dynambee said:

It would be a real shame to abandon such an advanced, high performance module at this point.

20 minutes ago, dynambee said:

Part of the agreement when the module was funded & created was that it would be supported by the TB team along with other TB-created modules.

Nobody is abandoning this module. If there's something broken in the module, it will be fixed. In the same way bugs are fixed in other modules. But that doesn't necessarily mean that new functionality will be implemented. For example, support for newer version of ES server. It would be nice to have, of course, but there's just no free capacity to do that. 

Now, what is wrong with the module that needs to be fixed? I haven't seen any new bug reports. If you did encounter any, please file github issue. Make sure you include reprosteps so we can reproduce the problem.

[vincentdenkspel]

41 minutes ago, datakick said:

but there's just no free capacity to do that. 

Maybe several people are interested to pay for compatibility. You never know.

[dynambee]

1 hour ago, datakick said:

Nobody is abandoning this module. If there's something broken in the module, it will be fixed. In the same way bugs are fixed in other modules. But that doesn't necessarily mean that new functionality will be implemented.

That is very good to hear. Thank you.
 

1 hour ago, datakick said:

But that doesn't necessarily mean that new functionality will be implemented. For example, support for newer version of ES server. It would be nice to have, of course, but there's just no free capacity to do that.

I don't have any problem with that. ES 5.4 is absolutely fine for index & search of any website up to 100s of thousands of products, and probably well beyond that. ES is capable of indexing & quickly searching vast amounts of data.
 

1 hour ago, datakick said:

Now, what is wrong with the module that needs to be fixed? I haven't seen any new bug reports. If you did encounter any, please file github issue. Make sure you include reprosteps so we can reproduce the problem

I'm making changes to my automation systems right now but once that is done (should be this week) I will be able to set up a decent sized test system to better check the performance and issues of the current module on 1.0.8 bleeding edge. It is difficult to properly test something like ES with only a small handful of products.

[movieseals]

1 hour ago, datakick said:

For example, support for newer version of ES server. It would be nice to have, of course, but there's just no free capacity to do that.

Well, that is the source of all my problems.  After speaking with tech support at my host, I finally learned that previous versions of ES, at least the free ones, are not secured.  The server will not accept anything below 6.8 because there is no security in the older versions unless it is the paid versions.  So it is impossible to make it work with the module at this point.  For the record, it would have been nice to know which versions of ES are supported in the documentation - it would have saved me a week of stumbling in the dark...

Edited June 18, 2019 by movieseals

[dynambee]

14 minutes ago, movieseals said:

Well, that is the source of all my problems.  After speaking with tech support at my host, I finally learned that previous versions of ES, at least the free ones, are not secured.  The server will not accept anything below 6.8 because there is no security in the older versions unless it is the paid versions.

The general way that ES is secured is by only allowing connections from localhost and not opening the firewall port to the outside world. In larger multi-server installations the ES server(s) resides on the local network and is not exposed to the outside world at all, all access comes from the web server.
 

15 minutes ago, movieseals said:

For the record, it would have been nice to know which versions of ES are supported in the documentation - it would have saved me a week of stumbling in the dark...

It is in the documentation, and I have been telling you from the beginning that the module is only known to be compatible with 5.4 and 6.0. Why you have ignored every one of my comment replies I do not know. @Factor even posted a link to the documentation in the thread you started about ES.

The documentation says, "Therefore, at the moment of writing, you will need a version of Elasticsearch that is: Version 5.4 or higher, including 6.0". At the time that the module (and documentation) was written 6.0 was the latest version of ES.

[eikichiz]

On 6/18/2019 at 3:48 PM, movieseals said:

Well, that is the source of all my problems.  After speaking with tech support at my host, I finally learned that previous versions of ES, at least the free ones, are not secured.  The server will not accept anything below 6.8 because there is no security in the older versions unless it is the paid versions.  So it is impossible to make it work with the module at this point.  For the record, it would have been nice to know which versions of ES are supported in the documentation - it would have saved me a week of stumbling in the dark...

There is a workaround for it, you could deploy a 5.4 ES server and do a ssh bridge or vpn so it won't be publicly exposed.

Edited September 1, 2019 by eikichiz