Grouper Developers Forum

[Chris Hyzer <>]

Hey,

 

I recently linked up Penn’s grouper diagnostics with our nagios monitoring, and then I had to resolve all unresolvable subjects and make sure no errors (which is a good thing), and now I will know if any loader job fails or if the grouper web services are down etc.  An enhancement for 1.6.1 is:

 

https://bugs.internet2.edu/jira/browse/GRP-469

 

allow WS status servlet to ensure certain groups have a minimum number of members

 

This is done. This is for sanity on your grouper deployment. i.e. that loader jobs dont remove all employees or something.

https://spaces.internet2.edu/display/GrouperWG/Grouper+diagnostics

You can assert that certain groups have a minimum number of members by any path (distinct)

Configure in grouper-ws.properties

#list groups which should check the size, in this case, "employee" or "students" in the key name is a variable
ws.diagnostic.checkGroupSize.employees.groupName = penn:community:employee
ws.diagnostic.checkGroupSize.employees.minSize = 15000

ws.diagnostic.checkGroupSize.students.groupName = penn:community:student
ws.diagnostic.checkGroupSize.students.minSize = 24000


This will cache the results (configurable), failures are cached for 5 minutes, successes are cached for 60 minutes.

Here is a snippet from the results:

SUCCESS minimumSize_penn:community:student: Found: 25184 members and expecting 24000 members (7457ms elapsed)
SUCCESS minimumSize_penn:community:employee: Found: 16574 members and expecting 15000 members (7556ms elapsed)


Here is a snippet from cache:

SUCCESS minimumSize_penn:community:student: Not checking, there was a success from before: 2010/07/30 12:08:10.501, expecting 24000, but had 25184 members. (21ms elapsed)
SUCCESS minimumSize_penn:community:employee: Not checking, there was a success from before: 2010/07/30 12:08:10.595, expecting 15000, but had 16574 members. (21ms elapsed)

 

Sample nagios config

 

Note: this will vary depending on your nagios setup.  Also, you can use any HTTP monitoring software, e.g. big brother, cron shell script, etc.

 

 #    /usr/local/nagios/libexec/check_http -I server.school.edu -S -u /grouperWs/status?diagnosticType=all -e "HTTP/1.1 200"

define service{

    use prodService

    service_description grouperWsAll production

    check_command check_http! -S -u /grouperWs/status?diagnosticType=all -e "HTTP/1.1 200"

    contact_groups  pennGroupsContacts

}

 

 

Thanks,

Chris