cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
618
Views
0
Helpful
2
Replies

CSS dns-record transitions

jbhurst
Level 1
Level 1

Hi,

I have a CSS11506 running 07.40.2.03s.

It has numerous DNS 'a' records configured, and a couple of them keep having transitions randomly as per below examples, but there are no associated service or content rule transitions that I can see.

(I've inserted xxx.xxx in place of actual name, and both .com and .co.uk are having the problem).

Anyone know what would be causing these dns-record transitions ?

Thanks

Jason

1 FEB 16:37:42 1/1 3520 NETMAN-4: dns-record: xxx.xxx.com state transition from UP to DOWN

1 FEB 16:37:42 1/1 3521 NETMAN-4: dns-record: xxx.xxx.co.uk state transition from UP to DOWN

1 FEB 16:37:46 1/1 3522 NETMAN-4: dns-record: xxx.xxx.com state transition from DOWN to UP

1 FEB 16:37:46 1/1 3523 NETMAN-4: dns-record: xxx.xxx.co.uk state transition from DOWN to UP

2 Replies 2

pgolding
Level 1
Level 1

Jason,

There has been one other report of this log message recently and in the end it turned out that the associated service was going from alive to dying state, then back to alive again. This indicates intermittent keepalive failures for the service, but not enough to take it down. For a service in dying state, the dns record will go down as you are seeing.

The resolution was to determine why the service sometimes did not respond to keepalives. A possible workaround is to alter the keepalive parameters for the appropriate service, to keep it in alive state. I would suggest using a sniffer to check the service response to keepallives.

Regards, Peter

greenberg.j
Level 1
Level 1

This can also happen if you do KAL-ICMP keepalive to your VIP

CSCtj38660 Bug Details

dns-record kal-icmp keepalive to vip could incorrectly go DOWN
Symptom:

The CSS is configured with a dns-record of type kal-icmp, querying a local VIP could incorrectly go DOWN when the service associated with that VIP is Alive.

Conditions:

The CSS is configured with a dns-record of type kal-icmp, querying a local VIP. The keepalive could incorrectly go DOWN if the service reports of a load of 254.

Workaround:

Configure "no load reporting" on the CSS