Forum Discussion

Piotr_Lewandows's avatar
Piotr_Lewandows
Icon for Altostratus rankAltostratus
Jun 05, 2017

FQDN node and icmp monitor

Hi,

I am just playing around with creating FQDN based nodes (v13.0.0HF2). I wonder what this error message means after standard icmp monitor is assigned to parent (non ephemeral) node:

Jun 5 10:31:52 bigiptest err bigd[13757]: 01060114:3: Cannot schedule :53 interval 0 timeout 1.

It's followed by this entries (and in GUI ephemeral nodes are marked with green dot)

Jun  5 10:31:53 bigiptest notice bigd[13757]: 01060143:5: Node /Common/lamp4-20_nd-10.1.20.14 address 10.1.20.14 session status enabled by monitor
Jun  5 10:31:53 bigiptest notice bigd[13757]: 01060141:5: Node /Common/lamp4-20_nd-10.1.20.14 monitor status up [ /Common/icmp: UP ] [ was checking for 0hr:7mins:12sec ]
Jun  5 10:31:53 bigiptest notice bigd[13757]: 01060143:5: Node /Common/lamp4-20_nd-10.1.20.16 address 10.1.20.16 session status enabled by monitor
Jun  5 10:31:53 bigiptest notice bigd[13757]: 01060141:5: Node /Common/lamp4-20_nd-10.1.20.16 monitor status up [ /Common/icmp: UP ] [ was checking for 0hr:2mins:7sec ]

Using standard icmp configuration:

ltm monitor icmp icmp {
    adaptive disabled
    interval 5
    time-until-up 0
    timeout 16
}

and interwal for querying DNS set to 60s - so much higher than interval and timeout in monitor

fqdn {
    address-family ipv4
    autopopulate enabled
    down-interval 5
    interval 60
    name lamp4-20.test.com
}

Is my config somehow broken? Should icmp monitor setting be set to higher timeout than fqdn interval?

Piotr

application delivery
autopopulate
devops
fqdn interval
fqdn node
icmp monitor
LTM

2 Replies

Sort By
  • dragonflymr's avatar
    dragonflymr
    Icon for Cirrostratus rankCirrostratus
    Jun 05, 2017

    I can't as well figure out how Down Interval works.

     

    According to manual

     

    In the Down Interval field, type the number of seconds the system waits to mark an FQDN node down following a DNS query failure.

     

    First of all what is considered DNS query failure?

     

    • No records returned for FQDN query
    • No reply from DNS server at all

    From my test it looks like status change for non ephemeral node are like that (same order as above events):

     

    • The DNS server(s) are available; No records returned 2017-06-05 09:59:00
    • The DNS server(s) are unavailable

    Now which node is set down by Down Interval? Non Ephemeral node or ephemeral node?

     

    Seems that first as only log entries are for non ephemeral node, for ephemeral I can't see anything.

     

    If ephemeral nodes has monitor assigned (via parent node) and there is reply from nodes there is situation when parent node is marked down (red diamond) but ephemeral up (green dot).

     

    A bit confusing situation.

     

    I guess if anyone noticed such issue:

     

    • Monitor assigned to parent node
    • Config saved
    • Entering node again - no entry in Active monitors - or rather there is entry but without monitor name - but monitor is working and listed when list ltm node is performed.

    I did some test for case when DNS is not reachable at all using tcpdump and ltm log.

     

    Down Interval set to 5s

     

    Results:

     

    13:23:14.723097 ARP, Request who-has 10.128.11.4 tell 10.128.11.10, length 35 out slot1/tmm0 lis= - last arp request for DNS IP

     

    Jun 5 13:23:25 bigiptest notice bigd[13757]: 01060141:5: Node /Common/lamp4-20_nd monitor status down [ 10.128.11.2: up ] [ was up for 0hr:2mins:21sec ]

     

    Jun 5 13:23:25 bigiptest notice bigd[13757]: 01060141:5: Node /Common/lamp4-20_nd monitor status down [ 10.128.11.4: down ] [ was up for 0hr:2mins:21sec ] - ltm log entries

     

    Form above it looks like non ephemeral node is going down after 11s - why so? Why not 5s?

     

    Piotr

     

  • dragonflymr's avatar
    dragonflymr
    Icon for Cirrostratus rankCirrostratus
    Jun 05, 2017

    I don't get what is relation between non ephemeral node status set by DNS query result and members availability.

     

    Scenario:

     

    • DNS not responding (after ephemeral nodes created)
    • Non ephemeral node with icmp monitor
    • members without monitor assigned

    Results:

     

    • Non ephemeral node is down (Offline (Enabled) - The DNS server(s) are unavailable)
    • Non ephemeral member is down (Offline (Enabled) - The parent node is down)
    • Ephemeral nodes are UP (icmp monitor)
    • Ephemeral nodes are Unknown
    • Pool is Unknown

    Traffic directed to VS using pool is handled without any issue - reaching backed servers

     

    Same situation when there is no monitor assigned to non ephemeral node.

     

    So what is the point of making non ephemeral node down because of DNS not replying when traffic is still processed by VS?

     

    Piotr