Elasticsearch not working properly with Chinese characters

[Sherry]

1. Search box in the top-right corner (the search box below the Settings) return no result.
2. Search box in the Pages can only locate the name of the pages (not searching content of the page).

Detail Information of server:
=======================
Elasticsearch server is running:
{
  "cluster_name" : "elasticsearch",
  "status" : "green",
  "timed_out" : false,
  "number_of_nodes" : 1,
  "number_of_data_nodes" : 1,
  "active_primary_shards" : 1,
  "active_shards" : 1,
  "relocating_shards" : 0,
  "initializing_shards" : 0,
  "unassigned_shards" : 0
}

The Elasticsearch plugin is currently active.
Host: 127.0.0.1
Elasticsearch port: 9200
Auth username: (not set)
Auth password: (not set)
Index name: mahara
Bypass index: (not set)
Elasticsearch analyzer: mahara_analyze
lasticsearch types: usr,interaction_instance,interaction_forum_post,group,view,artefact
Cron record limit: 0
Artefact types: Select all

Number of records of each type currently in the queue:
usr (1)
interaction_instance (0)
interaction_forum_post (0)
group (0)
view (0)
artefact (0)

- Mahara 1.10.1
- Ubuntu Linux 14.04.1
- Postgres database (version 9.3.5)
- Chrome Version 39.0.2171.95 m, Firefox 33.0.2, I.E. 11

[Kristina Hoeppner (kris-hoeppner)]

Do you have version 0.20 of Elasticsearch installed?

According to the documentation: "The plug-in was developed using version 0.20. It is not compatible with the latest version 0.90, as the PHP client Elastica v0.20.5.0 was written for Elasticsearch 0.20. " (https://wiki.mahara.org/index.php/Developer_Area/Specifications_in_Development/Elasticsearch)

[Sherry (cflfung)]

No, I don't have 0.20 Elasticsearch installed.
I can't find a way to install 0.20 Elastica and 0.20 Elasticsearch. Is there a setup guide for it?

[Aaron Wells (u-aaronw)]

Hi Sherry,

Sorry for the slow reply on this one. The wiki page Kristina linked to is actually somewhat out of date. The current version of Elasticsearch supported by Mahara is 0.90.1.0. That is a rather old version of Elasticsearch, and we have a ticket registered for updating it: https://bugs.launchpad.net/mahara/+bug/1384498

What version of Elasticsearch are you running?

Cheers,
Aaron

[Sherry (cflfung)]

Dear Aaron,

Thank you so much for your response. The version is 1.3.7. I recently find that it works with english search but fails with chinese search. Is there anything about system setting?

Best regards,
Sherry

[Aaron Wells (u-aaronw)]

Hi Sherry,

No, we haven't set up any specific system settings for Elasticsearch.

It's interesting that you mention it fails when you're searching for things in Chinese, though. On one of our client sites we've recently noticed a problem when Elasticsearch is trying to index student data that contains unusual Unicode characters. We still haven't figured out the exact cause. The Elasticsearch software itself supports these characters, and the Elastica library should support them, but for some reason it errors out.

Do you have any specific error messages from Elasticsearch in your Apache error logs?

Cheers,
Aaron

[Sherry (cflfung)]

Dear Aaron,

No, there is no error log for Elasticsearch.

Best regards,
Sherry

[Son Nguyen (ngson2000)]

Hi Sherry,

In Mahara 1.10, we use the PHP elasticsearch library client called Elastica version 0.90.2. This lib may not compatible to your elasticsearch 1.3.7.

Do you see any error in your cron log?