Bug #101
closedgenerate data set usage metadata/ provide access log
0%
Description
Tracking data set usage is an important part of running a metadata and data
archive. We need a new feature in the metacat server that tracks usage
statistics for each data-set and creates metadata records that document the
usage of data sets. This information is currently encoded in the
"data_request", "review_history", and "remarks" fields of eml-supplement.dtd.
These fields need to move forward as we revise these standards, and the metacat
servlet needs to automatically generate new records of this type as needed for a
data set, as it is accessed. We'll need to decide whether simply viewing the
metadata should be tracked, or whether a download is required to trigger a log
entry.
Related issues
Updated by Matt Jones almost 23 years ago
This information needs to be accessible to the data set owner (and possibly
others) through the Morpho interface and other interfaces.
Updated by Matt Jones almost 23 years ago
Moved issues that we do not intend to address during the current release to the
'Postpone' milestone.
Updated by Matt Jones over 21 years ago
- Bug 1104 has been marked as a duplicate of this bug. ***
Updated by Matt Jones over 21 years ago
Increasing priority and milestone because this issue has resurfaced a loit
recently, as evidenced by the recent duplicate bug #1104 that was filed.
Updated by Matt Jones over 20 years ago
Basic usage tracking is now implemented. Still need to test on oracle, but
basic functionality is working on postgres. Every time the user does one of the
following actions, it is logged in the access_log table, along with their uid,
ip number, the docid, and the date. I hav eyet to build a reporting funciton
for this, but it shouldn't be too difficult. The one problem I forsee is this
database growing too large too quickly, so one item might be a mechanism for
archiving the access_log table outside of the database on a regular basis. But
we'll see if that actually becomes an issue.
Still to do:
1) query/report functionality
2) make logging asynchronous to decrease the number of jdbc immediate calls
3) possibly write code to archive access_log data periodically
4) test on oracle
The access_log table records look like this, FYI:
entryid | ip_address | principal |
docid | event | date_logged
---------+---------------+--------------------------------------------+--------------------+--------+---------------------
5 | 24.237.20.168 | public |
knb-lter-gce.180.1 | read | 2004-04-02 12:31:46
6 | 24.237.20.168 | public |
obfs.45345.3 | read | 2004-04-02 12:32:19
7 | 24.237.20.168 | public |
obfs.45345.3 | read | 2004-04-02 12:32:36
8 | 24.237.20.168 | uid=jones,o=NCEAS,dc=ecoinformatics,dc=org |
obfs.45345.4 | update | 2004-04-02 12:33:41
9 | 24.237.20.168 | uid=jones,o=NCEAS,dc=ecoinformatics,dc=org |
obfs.45345.4 | read | 2004-04-02 12:35:05
10 | 24.237.20.168 | uid=jones,o=NCEAS,dc=ecoinformatics,dc=org |
obfs.45345.4 | delete | 2004-04-02 12:35:06
11 | 24.237.20.168 | public |
obfs.45346.1 | read | 2004-04-02 12:36:23
12 | 24.237.20.168 | uid=jones,o=NCEAS,dc=ecoinformatics,dc=org |
obfs.45350.1 | insert | 2004-04-02 12:39:24
13 | 24.237.20.168 | uid=jones,o=NCEAS,dc=ecoinformatics,dc=org |
knb-lter-gce.180.1 | read | 2004-04-02 12:45:08Updated by Matt Jones about 20 years ago
The query/report functionality (1) is now complete. Use the 'getlog' servlet
action to get an XML report of the events if you are in the adminsitrators group.
The function has been tested on oracle and is working so (4) is complete.
Have decided to indefinitely postpone asynchronous logging (2) and archiving of
log data (3) as its not clear that they are needed. Closing the bug as all
feattures are now complete.