Monitoring Backup Exec

[bill.unger]

I have upgraded to v1.5.2 and now see Database as an item type. Unfortunately, I cannot find any documentation on how to implement it. I see the script that NL_Dan has posted, but that doesn't seem to fit any parameters....

[NL_Daan]

Hi,
You need the following info for a database check:
DSN=<database source name>
user=<user name>
password=<password>
sql=<query>
the SQL query is

the query is:

SELECT * FROM Alert WHERE AlertDate > GETDATE() - 1 AND UserResponse='0' AND AlertTitle='Job Failed'

look on google for how to installing unixodbc

here are a couple of examples of unixodbc configurations for mssql and freetds

http://www.voip-info.org/wiki/view/FreeTDS

http://code.google.com/p/odbc-dbd/wiki/Linux

you'll need a /etc/odbcinst.ini and a /etc/odbc.ini
just make sure you got tcp enabled in sql express on the backup exec server

http://blogs.msdn.com/sqlexpress/archive/2005/05/05/415084.aspx

and create a user which you can use to authenticate (mixed authentication)

http://www.microsoft.com/downloads/details.aspx?FamilyID=c243a5ae-4bd1-4e3d-94b8-5a0f62bf7796&displaylang=en

[bill.unger]

look on google for how to installing unixodbc

Thanks for the all of the info!! Do I need unixodbc no matter what? For example, I am only going to be connecting to SQLServer on a remote Windows box...

thanks again!

[bill.unger]

I am still working through the implementation of this, have not installed the unixodbc, but have installed the 1.5.2 agent on the remote machine. I created the item (see below) but am getting a "Not Supported" message under Status on the Items Configuration page. I also have this line in the zabbix_server.log:

Parameter [db.odbc.select[BackupExec]] is not supported by agent on host [<myHost>] Old status [0]

Any thoughts or suggestions would be very much appreciated!

[NL_Daan]

Hi,
The odbc thingie in zabbix needs a dsn, dsn needs to be configured via unixobdc, so yes you'll need unixODBC

[Jason]

The event log monitoring seems to have settled down and now be quite stable. I have the logic right on the triggers and so far it has been quite reliable... I've posted the latest export of the template (also has other things in it)

I'm not sure the SQL method will work here as all of our remote windows servers are behind firewalls and running purely active checks.

[bill.unger]

Hi,
The odbc thingie in zabbix needs a dsn, dsn needs to be configured via unixobdc, so yes you'll need unixODBC

So does that imply that the zabbix server talks directly to SQLServer on the remote machine? Which would also imply that if the remote machine is on a different network behind a firewall, a hole must be poked through the firewall?

[sdwilders]

Monday - Friday backups

Hi,

I have backup monitoring working fine with an item that works across hosts that use a mixture of Backup Exec and NT Backup. The item I use is:

ZABBIX Agent (active)
eventlog[Application,ackup]
Log

I then use the following trigger:

(({Server - Backup:eventlog[Application,ackup].logseverity(4)}=4)&({Server - Backup:eventlog[Application,ackup].logsource(Backup Exec)}=1))|(({Server - Backup:eventlog[Application,ackup].logseverity(4)}=4)&({Server - Backup:eventlog[Application,ackup].logsource(NTBackup)}=1))

This all works fine but what I also want to check is if no backup jobs have run. I have setup the following:

({Server - Backup:eventlog[Application,ackup].nodata(90000)}=1)

This trigger fires if no backup items have been recorded in the Event log in the last 25 hours (gives the backup a bit longer than the previous in case the size of the backup has increased). The only problem is that some servers only backup Monday through Friday so on Sunday morning there are lots of triggers that don't reset until Monday night's backup has taken place. Is there a way to make nodata only apply to weekdays? Havin posed this question though, I am now thinking that as some backups complete in the early hours of the morning (so Fridays might complete on Saturday morning), would this effect such a trigger if it was possible?

Maybe the easiest way would be to set up a Saturday and Sunday backup.

PS - I use "ackup" as I seem to remember that this is case sensitive. Maybe this is why so many people have problems with event log monitoring?

[Jason]

Why not just use the dayofweek item in the triggers to exclude it on day 7 and 1 (assuming the monday backup runs late monday evening)?

That should work and stop any false triggers

[bill.unger]

How would one incorporate the dayofweek item into his pre-existing trigger code?

[Jason]

For example if his existing trigger is...

({Server - Backup:eventlog[Application,ackup].nodata(90000)}=1)

Then maybe something along the lines of...

({Server - Backup:eventlog[application].dayofweek(0)}#1&{Server - Backup:eventlog[application].dayofweek(0)}#7)&({Server - Backup:eventlog[Application,ackup].nodata(90000)}=1)

Or you could reference the dayofweek check to something such as system uptime or the like.

That is assuming that his backup starts monday evening.

Incidentally rather than checking for nodata from backup exec I'd look for the Job Start event instead as backup exec can make various event log entries from its housekeeping operations.