Statistics for Sakai and RULink
Sakai
Sakai Web Statistics
graphical web statistics
These are Standard Webalizer graphical reports.
"Hits" is every request to the server. For Sakai, even users who are idle are generating hits, because Sakai refreshes the list of users currently in the site every few seconds. These user list refreshes are counted in Hits and Files. They are not counted in Pages, and they are omitted from the "top URL" listing. (It appears that the number of user list refreshes is about the same as the number of Pages.)
"Files" is every request which actually transferred data (technically, it returned an HTTP code of 200, which means OK). This omits errors, cases where the client has a copy of the file that is still up to date, and a few utilities operations that don't transfer data.
"Pages" is intended to be actual HTML documents, as opposed to all the various images and support files. With Sakai it includes all the applications running in windows within the main page (technically, "iframes"). So if you refresh the whole page it will add several page counts: one for the main page and one for each subwindow. The window showing users currently in the site is not counted. It updates regularly, even when you haven't done anything, so if we counted it, it would inflate the page count unjustifiably.
"Visits" is based on the network (IP) address from which requests come. If requests from a given address stop for more than 30 minutes, it is assumed that the "visit" is finished. Further requests from that address start a new visit. This isn't perfect, but should be a reasonable guess.
"Sites" is the number of different network addresses from which we got requests. This is roughly the number of different computers that talked to us. However laptops will be assigned a different address if they move around, and in some cases a number of people share an address.
Performance information, every 10 min
ha1 - Java frontend
ha2 - Java frontend
ha3 - Java frontend
ha4 - Java frontend
ha5 - Java frontend
primary database server - Mysql
summary of all servers for last 30 min, refreshes automatically
- date
- percent CPU used by JVM/Mysql
- free memory [older: virtual memory size of JVM/Mysql]
- number of lightweight processes in JVM/Mysql
- number of connections to JVM/Mysql
- number of distinct hosts connected to JVM/Mysql
- number of active connections to Mysql [JVM only]
Usage Data
- date
- users with sessions in last day
- users with sessions in 7 days
- users with sessions in 30 days
- events since last entry
- K-12 users with sessions in last day
- K-12 users with sessions in 7 days
- K-12 users with sessions in 30 days
- guest users with sessions in last day
- guest users with sessions in 7 days
- guest users with sessions in 30 days
K-12 users includes teachers and students. Guests are primarily normal Rutgers faculty and students who were added to sites by email address rather than netid. However they can also be colleagues at other institutions.
K-12 and guests are included in the normal counts.
Events refers to entries in the Sakai events table. This includes all non-trivial actions by users. Normally this will be the number of events in the last day, but if for some reason the script doesn't run every day it will be the number of events since the last time the script ran. While most tools report events, note that Tests and Quizzes does not. So this count will not report usage of Tests and Quizzes.
This is historical data from the beginning of 2007 for the first and last columns of users. It goes through 9/27, because the users file will have it from then on. [I may reconstruct the whole users file.]
Portfolio statistics
Every day an entry is added. It contains a one-line summary and then a list of every site containing portfolios with a count of the number of portfolios.
The format of the one-line summary is
- date
- users who have created forms
- total number of forms
- users who have created portfolios
- total number of portfolios
Forms are listed because they show the number of people who have started working with a portfolio, but may not have gotten to the final stage where the actual portfolio is created.
For the portfolio site listing, numbers are users with at least one portfolio, and count of total portfolios.
Due Dates
To help us understand usage peak, we generate a list of due dates for the previous and next week. It is updated hourly.
RULink
calstats
date <sessions this day> <users over last 6 days>
mailstats
- date
- local deliveries. This includes deliveries to mailboxes, and mail discarded as spam, even if the mail would have been forwarded had it not been spam.
- local addresses. This is the number of unique local addresses delivered to. Mail to folders is counted as if it were delivered to the main inbox, so this is the number of users to whom mail was delivered.
- external deliveries. This is mail sent by our users to external addresses, and mail forwarded to external addresses. Before 10/5/2007 this number was much too high.
- forwarded addresses. This is the number of unique local addresses for which mail is forwarded. Note that before 10/5/2007 this number was much too high.
- spam. This is mail delivered to folders starting with "spam", and mail discarded. It is includes messages that would have been forwarded had they not been spam. Hence this count can be subtracted from local deliveries to get the local non-spam deliveries. Or you can add local and external deliveries, and use this to compute % spam. However you can't validly compute % spam individually for local or external mail, because the log doesn't have enough information to let me separate this.
- delay median, 90th percentile, and 99th percentile. That is, 50%, 90% or 99% of mail was delivered in this number of minutes or faster. The delay numbers are delays in minutes between arrival and actual delivery to a destination. We count only mail delivered to a local mailbox and mail forwarded to a destination within Rutgers. Mail forwarded outside Rutgers is subject to many influences outside our control. For mail with multiple addressees, the latest time is used. (Entries before 10/5/2007 counted only mail delivered locally.)
- forwarded. This is the number of messages forwarded. This number is included in the external deliveries count. If you subtract this from external deliveries, you'll get the number of messages sent by our users to external addresses. This number is new as of 10/5/2007.
imap
date <sessions this day> <users over last 6 days>
webmail
date <sessions this day> <users over last 6 days>
Disk space
Disk space on the NFS partition used by Sakai.
- time stampe
- file system
- total size bytes
- used bytes
- free bytes
- % free
- mount point
nfstest
date <time to copy 503434829 bytes>
This is a test of NFS performance. This has no direct relevance to users, but is used to make sure that a service that's critical to us is working. It should take about 40 sec
ldap
[Note: this is of historical interest. This ldap server is no longer being used in production]
date <sessions this day> <SSL sessions this day>
For more information, contact
rulink-support@rutgers.edu
©
2007
Rutgers, The State University of New Jersey. All rights reserved.