I have added some additional information to the SMTP Diagnostic test tool so that we can better troubleshoot Server Down alerts when customers are puzzled because they saw no outage.
I have added the full SMTP transaction transcripts to the SMTP Diag Alert emails. I also configured it to include the actual timeouts for “Timeout occurred due to inactivity” alerts which are what cause the SMTP Diag Failed emails.
Remember that just because we issue a SMTP failed does not mean that they were down or offline, just that they were unable to respond to our connection request within 15 seconds. They should have our testing networks whitelisted on their Firewalls and SMTP Defense mechanisms.
Thanks,
—
Peter LeBlond
Product Development Engineer
MxToolBox
Update 8/5/2010 11:55 CDT:We have made additional changes to the SMTP Monitoring code to try to alleviate the lingering false Down alerts. Please continue to give us feedback if you continue to experience them.
Peter:
As the mail server at 66.118.185.18 utilizes spam assassin which takes at least 20 seconds, I can understand why you monitoring alert occasionally sends information the system is down
The new code tries 2 times to get a connection. Also, we aren’t looking to complete the entire SMTP transaction in 15 seconds, just to establish a connection.
We are taking in the feedback from our users and checking our logs to try to find the best balance between performing the best for our users and having suitable performance for our servers which are trying to run a LOT of these.
Hello Peter,
thanks for this post. I love your tools and the website and want to give feedback to this comment:
1) In the Outage Email that tells me the server isnt avaible, I receive the following information about the transcript:
Full SMTP Transcript
2) The “back online” tells me the following:
Full SMTP Transcript
It looks like this isnt just working properly so far.
Regards,
John
to 1)
The blog didnt take the brakets and deleted the text inbetween. The mails tells me:
Full SMTP Transcript
Transcript Not Available
There is a descrepency between when the test & error occurs and when I got the notice (email). Your test says my server failed to respond on 8/19/10 @ 7:37:00P but I got the email at 8/19/10 @ 5:37P??
Brian,
My guess is that you live in the Pacific timezone. The notices are currently issued from Central time. We are looking at improving that aspect of the system by recording all times in UTC and then giving users a preference to set their own timezone for reporting.
Peter