ACE serverfarm total connections failures

cajalat · ‎10-31-2007

I have many server farms in various contexts and every single serverfarm shows something like this:

NDCD23ACERT1/Internal-1295# sh serverfarm FINPROD88_80 detail

serverfarm : FINPROD88_80, type: HOST

total rservers : 3

description : -

predictor : ROUNDROBIN

failaction : purge

total conn-dropcount : 0

---------------------------------

----------connections-----------

real weight state current total

---+---------------------+------+------------+----------+--------------------

rserver: FINPRODAPP5

10.1.129.39:8320 2 OUTOFSERVICE 2 138

total conn-failures : 348

rserver: FINPRODAW1

10.1.129.29:8320 1 OPERATIONAL 25 179

total conn-failures : 418

rserver: FINPRODAW2

10.1.129.30:8320 1 OPERATIONAL 8 165

total conn-failures : 456

The conn-failures counter keeps growning on all of the farms. What exactly could be the problem? People aren't complaining but it bothers me to see those counters grow.

Casey

Syed Iftekhar Ahmed · ‎10-31-2007

total conn-failures counter can increment due to several reasons

1. When Server sends RST to close the connection

2. When Client Sends an invalid HTTP request

3. If "persistence-rebalance" & "layer 7 class map" is configured and ACE doesnt find a L7 condition in client request.

So increments in the counter doesnt always means that there is a server problem :)

There are bugs reported for all the above conditions.I am not sure if they are fixed in the latest release.

Hope it helps

Syed

cajalat · ‎11-03-2007

I'm running A1(6.2) and the numbers are extremely suspicious. I don't think they are correct. I have 52 farms configured and every single one is showing a significantly large number of conn-failures. Generally around 50% of total connections.