For more information regarding the security incident at F5, the actions we are taking to address it, and our ongoing efforts to protect our customers, click here.

Forum Discussion

Benoit_C_'s avatar
Benoit_C_
Jul 15, 2013

interval/up-interval vs timeout

Hello,

 

 

As mentionned in the summary, I would like to get some explanations on how the 'timeout' for an HTTP monitor will react when the up-interval value is set to a different value than the 'interval' one.

 

Iin my case, up-interval is bigger than interval.

 

 

 

I think I pretty get the story of the 3:1 ratio between timeout and interval, it helps to determine how many consecutive request must fail before declaring a server as down.

 

Interval of 5 and timeout of 16 means that after three failed checks, the resource is declared as down.

 

 

I am currently using values of 30sec for the interval (frequency) and 62 for the timeout (OK, not really a 3:1 ratio but proven to work)

 

Meaning that we wait for two consecutives failed HTTP checks before declaring a server as down.

 

 

My client asks me if we could keep 30 sec as interval when the server is up, but use a more 'aggressive' interval (5 sec in fact) in case of no reply from a server (within the timeout period).

 

I think to configure for this the 'up-interval' to 30sec, the (down) 'interval' to 5.

 

 

But I am not sure if "(down) interval" already applies when a member is failing (but still marked as up) or only after when marked as down ?

 

(I do not really have a test bed to play with values, any change occurs in our 'prod' environment)

 

 

Related to 1st question, I am wondering if, as for the "up-interval" and "(down) interval", there is a way to use two different timeout values: one when a member is up or down, one when a member is in a 'failing' state but not already declared as down ?

 

 

thanks and best regards,

 

 

--

 

Benoit

 

 

 

 

 

management
monitoring

3 Replies

  • nitass's avatar
    nitass
    Icon for Employee rankEmployee
    Jul 15, 2013
    But I am not sure if "(down) interval" already applies when a member is failing (but still marked as up) or only after when marked as down ?it is applied after pool member is marked down. 

    root@(ve11a)(cfg-sync Changes Pending)(Active)(/Common)(tmos) list ltm pool foo
ltm pool foo {
    members {
        200.200.200.101:80 {
            address 200.200.200.101
            session monitor-enabled
            state checking
        }
    }
    monitor mytcp
}
root@(ve11a)(cfg-sync Changes Pending)(Active)(/Common)(tmos) list ltm monitor tcp mytcp
ltm monitor tcp mytcp {
    defaults-from tcp
    destination *:*
    interval 5
    time-until-up 0
    timeout 62
    up-interval 30
}

 httpd was shutted down at 03:36:25. pool member was marked down at 03:37:57. up-interval started working after the monitor marked pool member down.  

[root@ve11a:Active:Changes Pending] config  tcpdump -nni 0.0 host 200.200.200.101 and port 80 and 'tcp[13] = 2'
tcpdump: verbose output suppressed, use -v or -vv for full protocol decode
listening on 0.0, link-type EN10MB (Ethernet), capture size 96 bytes
03:35:25.030584 IP 200.200.200.11.52748 > 200.200.200.101.80: S 3739229516:3739229516(0) win 14600 
03:35:55.077977 IP 200.200.200.11.52755 > 200.200.200.101.80: S 2385607889:2385607889(0) win 14600 
03:36:25.026322 IP 200.200.200.11.52762 > 200.200.200.101.80: S 1365983269:1365983269(0) win 14600 
03:36:55.072981 IP 200.200.200.11.52779 > 200.200.200.101.80: S 1968263475:1968263475(0) win 14600 
03:37:25.021490 IP 200.200.200.11.52785 > 200.200.200.101.80: S 1320259441:1320259441(0) win 14600 
03:37:55.072075 IP 200.200.200.11.52792 > 200.200.200.101.80: S 2067005880:2067005880(0) win 14600 
03:38:00.078257 IP 200.200.200.11.52794 > 200.200.200.101.80: S 1220730320:1220730320(0) win 14600 
03:38:05.086906 IP 200.200.200.11.52796 > 200.200.200.101.80: S 2381723492:2381723492(0) win 14600 
03:38:10.094225 IP 200.200.200.11.52798 > 200.200.200.101.80: S 96810804:96810804(0) win 14600 
03:38:15.100082 IP 200.200.200.11.52800 > 200.200.200.101.80: S 199539598:199539598(0) win 14600
  • Benoit_C_'s avatar
    Benoit_C_
    Jul 22, 2013
    Hi,

     

     

    thanks for the reply.

     

    And I confirm the LTM is smart enough to handle an 'up interval' with a value higher than the one configured in 'timeout'. As soon as '(down) interval' remains lower than 'timeout'.

     

     

    Anyway, my feeling is that the wording used for keywords is not always obvious to get :-(

     

     

    hope this helps

     

    best regards,

     

     

    --

     

    Benoit
  • Brian_Mayer_841's avatar
    Brian_Mayer_841
    Icon for Nimbostratus rankNimbostratus
    Feb 08, 2014

    Hello there,

     

    I want to do the opposite, as described here:

     

    https://devcentral.f5.com/questions/how-to-reduce-health-monitor-interval-when-pool-member-is-downanswer88455

     

    Eseentially, I want my up interval to always be 5s, with a [down] interval of 30s. The timeout would be unchanged I guess? However, I should note that, in this config, the standard timeout to interval ratio (3n+1) is violated, since in reality, the typical interval can be thought of as the 'up interval' rather than the actual 'interval' field; and this'interval' value (in my case, 30s) would only ever be enforced if the pool member is down.

     

    Is this logic sound and, if so, I guess I can give it a go in our Test environment? Just wanted to hear what you guys had to say about it before I just plug and play!

     

    Thanks! B