User Activity Log Info
| Project: | Drupal.org infrastructure |
| Component: | Other |
| Category: | task |
| Priority: | normal |
| Assigned: | Unassigned |
| Status: | active |
Jump to:
I'm working on another survey for the doc team.
Where would I find what user activity is logged on the drupal.org site?
How long is the activity log kept?
What info is noted when the activity occurs?
What's the process to get something else added to the activity log?
I know there are many things that folks do, but I'm more interested in learning what info is available that might help me with documentation.
For example, how many pages a week/month/qtr/year (some timeframe) are deleted, changed, added.
Are the folks that do the deletes the same ones that do adds or changes OR are there really 3 separate kinds of contributors?
If a person does a change, how long on average were they members of drupal.org before they started helping with docs, are they generally from a particular country?
Is there a particular type of person that puts the most comments on pages (determined by how long they've been a drupal member).
Things like that.
I don't have the survey figured out yet, partly because I don't know what is available currently and what I'd need to provide a patch for. Any guidance would be appreciated.
Thanks,
bekasu
#1
> Where would I find what user activity is logged on the drupal.org site?
Probably someone with access to the DB could run a report for you. It is basically the "normal" things that Drupal logs. There are no additional logging modules installed.
> How long is the activity log kept?
We keep 500,000 rows in the database. I don't see additional (e.g. syslog) logging locations, but maybe this is handled via the DB somewhere. Practically speaking, 500,000 records means 08/06/2009 - 02:11 so it seems we keep data for less than 48 hours.
> What info is noted when the activity occurs?
Standard "watchdog" stuff. Hostname, username, action.
> What's the process to get something else added to the activity log?
Request it in the infrastructure issue queue (or more accurately in http://drupal.org/project/issues/drupalorg but that tends to get a lot fewer people looking at it than infrastructure so even though infra is off-topic, it can be the best place to request).
#2
We actually do log extra stuff with the role_activity module:
http://cvs.drupal.org/viewvc.py/drupal/contributions/modules/drupalorg/r...
This is mostly because watchdog expires quite quickly.
#3
Added to what greggles said, we also have:
- additional logging for the activity of privileged users (administrator, site maintainer, CVS administrator, documentation maintainer, user administrator). Node, comment and user creation/edition/deletion are logged, in addition to user login and logout. I'm not sure we have a retention policy, but the current log goes back to mid-February.
- the standard Apache/Squid access log on all the web nodes. Not sure about our retention policy on those either.
#4
is the role_activity module only available via cvs? I couldn't seem to find it in downloads or over on drupalmodules.com ..
if I pull down the cvs code, is the 'extra stuff' already activated or would I need someone to tell me what 'extra stuff' is currently active?
I'm not interested in login attempts, logout attempts, password resets, or that type of thing.
I'm only interested in node changes/deletes/adds, time spent changing something, dates nodes were modified, that type of thing
#5
Damien,
Do I need to get add1sun to request access to log data for me OR should I provide the info to add1sun for her to ask? I know she is on the road pretty seriously this summer; however, if you tell me what I need to do I can write it up and send it through the proper channels.
I don't mind doing the datamining, but I don't want to overstep my limits in the drupal community either.
Once I get a handle on how frequently things get changed, what types of changes occur, when changes spike, etc. I would like to merge the information back with the id profile info to get the time they've been a drupal.org member and perhaps what country they are from.
I'm not sure that country is relevant; however, we have so few demographics available that I hesitate to eliminate it from the analysis until I've run basic analysis.
A slice of log data would be fine, whatever timeframe that would be convienent and in whatever format of the data although it would be easier if it were a standardized format rather than variable format. Since I'm begging, I'll take whatever I can get.
Also, do I need to provide/sign a non-disclosure agreement so you folks have a comfort level that I'm not planning on harvesting info for some nefarious purpose?
Again, forgive me for my ignorance but I just want to be sure I'm doing things appropriately.
Thanks in advance & I do appreciate the detailed response.