Forum Discussion
mpete32_168869
Cirrus
CirrusFailure of SMTP health check
I've always received great help here and it's truly appreciated. Now I have another issue that requires some assistance. BTW, I have a support case with F5 regarding this and they have not been able to offer any solutions that have worked.
I have a virtual server that uses a single member pool doing SMTP health checks. The health check is failing (showing the member down) but all logs show proper connectivity, proper smtp responses from the mail server, and communications working correctly, but the health check still shows down and the logs related to this traffic are displaying a "Sending data failed, errno 'EPIPE'" type error message. I can telnet from the F5 to the member server from a SSH console and run the same SMTP commands and all checks out. Simply not sure why the health check would fail is all appears to be talking properly.
@mpete32
You can also try to re-create SMTP monitor. Set Interval to 30, and Timeout to 91. And enable debug (Advanced configuration for the monitor), it may show some info in logs...
JGCumulonimbus
Which version of BIG-IP are you running? Did the service work before? Do you have any non-default route domain configured? Do you have any SNAT configured for the service? What is your smtp monitor like?
nitassEmployee
have you tried tcpdump? could it be helpful?
tcpdump -nni 0.0:nnn -s0 -w /var/tmp/output.pcap host x.x.x.x and host y.y.y.y and port zzz -v x.x.x.x is non-floating self ip on pool member vlan y.y.y.y is pool member ip zzz is pool member port
Dmitri_Ch__1425@mpete32
You can also try to re-create SMTP monitor. Set Interval to 30, and Timeout to 91. And enable debug (Advanced configuration for the monitor), it may show some info in logs...
mpete32_168869I am using the latest version for BigIP LTM which is 11.6.0 Build 1.0.403. This is the first SMTP health check and VIP that I'm setting up on the appliance. SNAT is set for Automap and the SMTP monitor has been set at 5/16 and 30/91 for interval times with no difference. I even set up a TCP monitor checking ort 25 and it checks good. As Dmitri suggested before, I enable the Debug logs for the SMTP monitor and the logs show that the server is responding to the SMTP commands. Very odd. Everything shows proper connectivity and communication but the health check simply won't pass when its set to SMTP. it may be a flaw in the version I'm using. I'm going to step back a version and try again.- Dmitri_Ch__1425I'm on "BIG-IP 11.6.0 Build 1.0.403 Hotfix HF1" as well. SMTP monitor reports 'green'
- mpete32_168869
Included an image of the tcpdump capture in Wireshark.
- nitassjust wondering why it took about 28 seconds to send smtp helo command after 3 ways handshake.
- Dmitri_Ch__1425
@mpete32,
Which monitor does fail in your case? Node or Pool? Your node monitor may be failing which marks Pool as 'down', even SMTP checks are OK. I.e. default ICMP monitor doesn't reach SMTP server. You can try creating TCP half open for port 25 and assign it to the node.
- mpete32_168869
I had to increase the time out for the health check from 30/90 to 30/120. That seemed to help with the issue of the delay in the responses between the LTM and the member server.
- nitassi never used smtp monitor but i think the delay is not normal. increasing interval/timeout will affect how fast bigd detect pool member down.
- Khamari_2736
Nimbostratus
I'm having the same exact issue, with 11.6 HF3.36 SMTP worked on 11.3 HF9. increasing the health check from 30/90 to 30/120 did help in my case!
- Khamari_2736
I meant didn't help in my case!