Forum Discussion
dimka___104021
Nimbostratus
NimbostratusLB_FAILED, but pool is UP
Hi,
I'm terminating SSL on F5 and sending http requests to backend
Weblogic pool. In this pool i have only 1 member.
This part of irule logs failed pool information:
when LB_FAILED {
log local0. "[LB::server pool]:[LB::server addr]:[LB::server port] - status: [LB::status]"
}
In /var/log/ltm there're a lot of LB_FAILED errors.
But LB::status is always "up".
Jan 30 18:07:35 tmm tmm[13018]: Rule vs_1_irule : wl_pool:192.168.88.238:8000 - status: up
MSIE browsers show error "Page cannot be displayed". When i
change in TCP profile "Maximum Syn Retransmissions" to 12
(default is 4) all errors are gone. Everything works as it should.
Why LB_FAILED is triggered if LB::status is "UP"?
What does "max syn retransmission" mean and how high
number can affect performance?
1 Reply
dimka___104021Hi,
I'm using http health monitor for pool with Interval=5 and Timeout=16 seconds.
Same backend is used by our current iPlanet webserver and there're no issues/errors for clients.
Through F5 i'm passing couple percents of whole traffic (approx. 100-150 active connections).
tcpdump shows a lot of retransmission from F5 to backend:
13:20:44.741601 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
13:20:45.941601 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
13:20:47.141602 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
13:20:49.341602 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
13:20:53.541601 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
13:21:01.741601 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
13:21:17.941601 193.40.195.59.1867 > hnet.crebit.ee.8000: S 3215066501:3215066501(0) win 4380 (DF)
p.s. And there're no errors on F5 interfaces
