Few questions about Enhanced Search

[Nuno]

Hi

I don't have ES yet and was just playing with ElasticSearch to understand how it works and what resources does it need.
I have a relative small board with just 350k posts and 100k images.

• Does stemming works with Portuguese European language?
• How do you determine the optimal value for -Xms and -Xmx?
• Any downside using java-1.8.0-openjdk-headless and elasticsearch-oss instead of java-1.8.0-openjdk and elasticsearch in centos 7?
• How does the 12 month support will match the XF renewal date from December if I get ES now?

Thanks

 

[Mike]

Does stemming works with Portuguese European language?

There is a stemming option for Portuguese.

How do you determine the optimal value for -Xms and -Xmx?

Ideally, you'd want to keep the full index in memory, though with that size of forum, it shouldn't be difficult. I believe, on average, 1 million documents comes out around 300MB in reported size, though the memory usage might be a bit different. Those options setup the min and max memory limits.

(I can't really speak to differences between those package manager packages.)

How does the 12 month support will match the XF renewal date from December if I get ES now?

Yes, the extension date is always based on the main XF license.

 

[Nuno]

Ideally, you'd want to keep the full index in memory, though with that size of forum, it shouldn't be difficult. I believe, on average, 1 million documents comes out around 300MB in reported size, though the memory usage might be a bit different. Those options setup the min and max memory limits.

So for 447,396 (90.6 MB) Documents, 128M should be more tan enough?

 

[Mike]

I suspect you may well be cutting it thin as I'm not exactly positive on the how the data is represented in memory, how it handles temporarily result sets, etc. There are a lot of variables here as it would vary quite a bit on your server specs, the other processes running on your server and how they're configured, whether it's a spinning HD or SSD, etc.

I would probably allocate more memory to avoid potential issues (especially as that index will get bigger over time).

 

[Nuno]

This are the stats I have now.
I think I have some space left to grow, don't I?
This is with Xms64m and Xmx64m.

 

[Nuno]

@Mike do you use some of the features from X-Pack or just the barebones storing and indexing?

 

[Mike]

Just Elasticsearch itself.