Forum Discussion

lipos_54863's avatar
lipos_54863
Icon for Nimbostratus rankNimbostratus
Aug 31, 2010

effective single node persistence

Hi all,

 

 

I'm looking for the most efficient way of solving my problem.

 

I basically need a single node persistence functionality for on of mine VS.

 

The "Single Node Persistence" iRule is exactly what I need.

 

The question is what the advantage of using it over the build-in "Destination Persistence"? Is it exactly the same functionality with different implementations?

 

 

Additionally, I need to know if the DB1 will failover to DB2.

 

This can be achieved by log entry in the "Single Node Persistence" iRule.

 

The problem is that I need something more - ideally an email notification.

 

 

I know that I can change the configuration of the F5s so that I will be able to generate an email based on an SNMPTrap generated under the iRule.

 

The problem is that I don't want to change the default F5s configuration since I don't have any test environment that I can test it.

 

 

I know that I can analyse the external syslog and generate an email based on it, but still I can't afford have a weak link as some syslog/script system failing affecting this my reporting mechanism.

 

 

iConrol is out of the question as well.

 

 

The perfect solution for me would be to query SNMP on the F5s using my external monitoring system.

 

 

The question is what I would monitor?

 

I'm not sure if LTM are able to provide me this kind of information's through SNMP.

 

Ideally I would like to monitor the persistence table, or create simple condition in the iRule that will populate SNMP variable so I would be able to monitor it using my SNMP monitor. The only thing that I would need is an OID of this iRule variable.

 

 

The second functionality under the iRule would be really powerful, but is it available?

 

In details, are variables from an iRule are able to be monitored using external SNMP queries.

 

 

This mechanism gives me a possibility of monitoring the DB1 and DB2 failover and the monitoring mechanism it self a since it will become critical.

 

 

Let me know if you have any suggestion.

 

Thanks!

 

3 Replies

  • Number of connections on node basis is doing it's thing but still maybe something more precise than that?

     

    snmpwalk -c public -v2c 192.168.0.1 -On .1.3.6.1.4.1.3375.2.2.4.2.3.1.9
  • Hi Lipos,

     

    Here is a shell command that could work for you

     

     

     

     

    b conn | grep | awk '{print $1 }' | cut -d : -f 1 | uniq -c | sort

     

     

     

    I hope this helps

     

     

    Bhattman
  • Thanks but that's taking too long.

     

     

    I think that my solution should work for my monitoring system.