Difference between revisions of "Stats Design"

From Dreamwidth Notes
Jump to: navigation, search
m (added needs-update flag)
 
Line 2: Line 2:
  
 
This is a a loosely organized collection of notes and links for the Stats feature design. (Note: this is for the new stats system only - there's an existing legacy system too.)
 
This is a a loosely organized collection of notes and links for the Stats feature design. (Note: this is for the new stats system only - there's an existing legacy system too.)
 
[http://bugs.dwscoalition.org/show_bug.cgi?id=2717 Bug 2717] and its dependent bugs
 
  
 
[[Project_Teams#Site_statistics|Stats project team]]
 
[[Project_Teams#Site_statistics|Stats project team]]

Latest revision as of 10:08, 4 August 2014

Needs Update: This page has been found in need of an update. Information may be wrong or outdated!

This is a a loosely organized collection of notes and links for the Stats feature design. (Note: this is for the new stats system only - there's an existing legacy system too.)

Stats project team

Table definitions

CREATE TABLE statkeylist (
    statkeyid INT UNSIGNED NOT NULL AUTO_INCREMENT, 
    name VARCHAR(255) DEFAULT NULL, 
 
    PRIMARY KEY (statkeyid), 
    UNIQUE KEY (name) 
)
 
CREATE TABLE site_stats ( 
    category_id INT UNSIGNED NOT NULL, 
    key_id INT UNSIGNED NOT NULL, 
    insert_time INT UNSIGNED NOT NULL, 
    VALUE INT UNSIGNED NOT NULL, 
 
    -- FIXME: This is good for retrieving data for a single category+key, but 
    -- maybe not as good if we want all keys for the category, with a limit on 
    -- time (ie, last 5 entries, or last 2 weeks). Do we need an extra index? 
    INDEX (category_id, key_id, insert_time) 
)

Storing and retrieving statistics

DW::StatStore

Statistics items

Collecting statistics

This works by massaging, aggregating, and or compiling (as appropriate) data from various places and storing the results using DW::StatStore.

1. etc/stats-collection.conf should contain something like:

## collect and store all stats under the accounts category
## hangs if you don't enclose the asterisk in quotes
# accounts: "*"
## collect and store only these stats under the accounts category
# accounts: [ identity, personal, redirect ]
accounts: "*"
active: "*"
paid: "*"

2. bin/worker/stats-collection is the utility that handles stats collection. Run at least once per 24 hours, either from your crontab or by listing it in etc/workers.conf.

3. (only needed if you chose to use etc/workers.conf): add a line to the "all" section reading:

  stats-collection: 1
(note: this is for development only, as it assumes you only have one server - in production environments, you will usually list it in the section for a specific server)

Displaying statistics