DX NetOps

 View Only
  • 1.  BIGIP Service Down alert filtering

    Posted Jan 11, 2017 10:48 AM

    I have successfully configured a SANM policy filter for BIGIP SERVICE DOWN and BIGIP NODE DOWN alerts, which are essentially traps received for servers either failing a health probe or administratively being brought offline in the LTM.  That's all well and good, but it looks to me like I can't filter down any more granular than the device level.  One LTM may contain virtuals/pools that are managed by different Operations groups, which means they would only be interested in alerts for the pools they are directly responsible for.  Does anyone have any idea how I could further filter based on contents in the payload of the trap as opposed to just the alarm type for the whole device?  Thanks.



  • 2.  Re: BIGIP Service Down alert filtering

    Broadcom Employee
    Posted Jan 11, 2017 10:55 AM

    One suggestion would be to use event rules to generate unique alarms based on the trap payload. 



  • 3.  Re: BIGIP Service Down alert filtering

    Posted Jan 11, 2017 11:40 AM

    That's a fantastic idea.  My experience with event rules is fairly limited though.  Would anyone happen to have an example of what the rule would look like for event 0x3b70012 based on a string in variable {S 2}?



  • 4.  Re: BIGIP Service Down alert filtering
    Best Answer

    Broadcom Employee
    Posted Jan 11, 2017 12:26 PM

    It's not exact but I did a similar exercise recently for Citrix Netscaler trap where the event is 0x5b80007 and the interesting varbind is 1 instead of 2.  Anyway, I copy that event off to new ones (0xfff00001, 0xfff00002, and 0xfff00003) that I want to trigger based on varbind contents, then modify the original event to not generate an alarm right away but rather do some event processing:

     

     

    The first two events are my "unique" ones that match a certain regular expression and the last one is a default which is just the original event/alarm.  Here's the rule for the first one:

     

     

    Or if you just wanted to see the EventDisp:

     

    0x5b80007 E 20 R { 1 } CA.EventCondition, "(regexp({v 1}, {S \"_lbvs\"}))" , "0xfff00001 1:1,2:2,1:76620","(regexp({v 1}, {S \"_gss\"}))" , "0xfff00002 1:1,2:2,1:76620","(default)" , "0xfff00003 1:1,2:2,1:76620"
    0xfff00003 E 20 A 1,0x5b80007,1
    0xfff00002 E 20 A 3,0xfff00002,1
    0xfff00001 E 20 A 3,0xfff00001,1

     

    I did something extra where I copied the value of my "interesting" varbind 1 to event variable 76620 because we wanted to see that as the alarm title and not the default "ENTITY DOWN":

     

     

    I don't know what is in those varbinds from BIGIP to know if that would be helpful to you.  Also, if we didn't want different probable cause text for the 01 and 02 events, they all could have used the orginal 0x5b80007 PCause that the default 03 event uses.

     

    Hope this helps.

     

    -Rob



  • 5.  Re: BIGIP Service Down alert filtering

    Posted Jan 11, 2017 04:41 PM

    That is a fantastic example Rob and should give us enough to work with.  Thank you!