Open Hywan opened 8 years ago
@K-Phoen is responsible to lead this project.
Hi,
actually, we have 141 documents indexed in es.
disk usage: 172MB (nGram tokenizer (3-20)) memory usage: 142MB (heap)
@K-Phoen if you want, I can give you an access to Elastic HQ, where you could found all stats about the current instance.
It can be useful, maybe, don't know: http://blog.adrien-gallou.fr/2015/09/21/pertinence-elasticsearch.html, by @agallou.
I already read this great article ! But, it targets only the relevance, nothing about the memory usage.
And, I said you, you should not have a problem about memory with 150 articles.
Same here, I don't think that the number of documents will be a problem but if needed we'll keep an eye on the global memory consumption (documents + server itself).
I'll try to replicate on my machine the elasticsearch setup used by Hoa (with the same documents) to experiment on the memory usage and to have numbers to give @Hywan and @Pierozi. This should help them to determine their needs in "server sizing".
Thanks @K-Phoen that should be helpful because i didn't' know much about elasticsearch
consumption
@GuillaumeDievart Could you put here the URLs of the search code repository please?
Hi @K-Phoen,
you have just to clone this repo https://github.com/GuillaumeDievart/Hoa-Search and to create the analyzer via the cmd Data/Bin/Init.php, then you can run the crawler.
Careful, this configuration has been created in May 2014, there are maybe few change since (on the es version).
I think, you can check the analyzer and mapping (auto, without boost, adjust analyzer by language ...), if you want to improve the relevance.
@GuillaumeDievart Hope you understand that we would like to get eveything packed in the same place? We still want you on board!
Yes, I understood,
"We still want you on board!"
@Hywan I hope !! ;)
@GuillaumeDievart awesome! I'll play with it when I have some free time :)
@K-Phoen Did you got time to check a little bit?
Our search service is based on ElasticSearch. This application likes to eat memory. We need to optimize it.
Original author of ES instance: @GuillaumeDievart.
Tasks