edmund@qualcom.qualcomm.com (The Silver Surfer) (03/10/90)
I seem to be having problems having sendmail.5.61 deferring mail, when
mail is sent to a local host that is down. The sendmail that I am using
now (sendmail.4.0), doesn't seem to have this problem.
I tried updating the sendmail.cf file using the m4 utilities in cf/cf for
a main machine, and still it didn't work. When I use the old sendmail using
the new sendmail.cf file, the results are the same. It defers the mail like it's
supposed to.
This is what it looks like, sending mail to cancun (who's down) using
sendmail.4.0 in verbose mode:
edmund@cancun... Connecting to cancun.qualcomm.com via ETHERNET...
Trying 129.46.4.6... Connection timed out during user open with cancun.qualcomm.com
edmund@cancun... Deferred: Host cancun.qualcomm.com is down
The mailq looks like this:
Mail Queue (1 request)
--QID-- --Size-- -----Q-Time----- ------------Sender/Recipient------------
AA20315 1 Tue Mar 6 09:50 edmund
(Deferred: Host cancun.qualcomm.com is down)
edmund@cancun
This is what senmdail.5.61 looks like in verbose mode with the same .cf file:
edmund@cancun... Connecting to cancun.qualcomm.com (tcpld)...
220 qualcom.qualcomm.com Sendmail 5.61-bcr/1.34 ready at Tue, 6 Mar 90 10:01:12 -0800
>>> HELO qualcom.qualcomm.com
553 Local configuration error, hostname not recognized as local
>>> QUIT
221 qualcom.qualcomm.com closing connection
edmund@cancun... Service unavailable: Bad file number
Saving message in /usr2/edmund/dead.letter
/usr2/edmund/dead.letter... Sent
The mailq is empty as well. Any ideas what it could be?andy@jhunix.HCF.JHU.EDU (Andy S Poling) (03/12/90)
In article <1894@qualcom.qualcomm.com> edmund@qualcom.qualcomm.com (The Silver Surfer) writes: >I seem to be having problems having sendmail.5.61 deferring mail, when >mail is sent to a local host that is down. The sendmail that I am using >now (sendmail.4.0), doesn't seem to have this problem. [...] >This is what it looks like, sending mail to cancun (who's down) using >sendmail.4.0 in verbose mode: > >edmund@cancun... Connecting to cancun.qualcomm.com via ETHERNET... >Trying 129.46.4.6... Connection timed out during user open with cancun.qualcomm.com >edmund@cancun... Deferred: Host cancun.qualcomm.com is down [...] >This is what senmdail.5.61 looks like in verbose mode with the same .cf file: > >edmund@cancun... Connecting to cancun.qualcomm.com (tcpld)... >220 qualcom.qualcomm.com Sendmail 5.61-bcr/1.34 ready at Tue, 6 Mar 90 10:01:12 -0800 >>>> HELO qualcom.qualcomm.com >553 Local configuration error, hostname not recognized as local This looks like you're running into MX problems. Sendmial 5.61 doesn't (IMHO) handle very well the situation where it is the "secondary" MX for a host that is down. It will connect to itself and cause the "configuration error". The mail in question then gets bounced back with this very un-helpful error message. I think that if all MXs of lower preference are un-reachable, sendmail should re-queue the mail and retry on the next queue run. I began to implement this behavior, but never finished the job. You don't have this problem with sendmail 4.* because it doesn't use MXs. -Andy -- Andy Poling Internet: andy@gollum.hcf.jhu.edu Network Services Group Bitnet: ANDY@JHUVMS Homewood Academic Computing Voice: (301)338-8096 Johns Hopkins University UUCP: mimsy!aplcen!jhunix!gollum!andy
towfiq@interlan.Interlan.COM (Mark Towfigh) (03/18/90)
In article <1894@qualcom.qualcomm.com> edmund@qualcom.qualcomm.com (The Silver Surfer) writes: I seem to be having problems having sendmail.5.61 deferring mail, when mail is sent to a local host that is down. The sendmail that I am using now (sendmail.4.0), doesn't seem to have this problem. [illustration of how the 4.0 version works] This is what senmdail.5.61 looks like in verbose mode with the same .cf file: edmund@cancun... Connecting to cancun.qualcomm.com (tcpld)... 220 qualcom.qualcomm.com Sendmail 5.61-bcr/1.34 ready at Tue, 6 Mar 90 10:01:12 -0800 >>> HELO qualcom.qualcomm.com 553 Local configuration error, hostname not recognized as local >>> QUIT 221 qualcom.qualcomm.com closing connection edmund@cancun... Service unavailable: Bad file number Saving message in /usr2/edmund/dead.letter /usr2/edmund/dead.letter... Sent The mailq is empty as well. Any ideas what it could be? OK - this is a CF file problem. Notice in the transaction above that your machine, qualcom, is trying to connect to itself to deliver mail to cancun. That's bad, because it means that qualcom has failed to realize that it *is* qualcom -- if Sendmail didn't pick this up, we would have a mail loop. Without knowing how your CF file is set up, it's tricky to know what's going on. Actually I take this back -- it seems, from your above script, that qualcomm thinks it's connecting to cancun, and that it's surprised when it winds up talking to itself. Could it be that the host information for your site is not entirely correct? I'll bet it's a /etc/hosts file vs. BIND discrepancy. I'll bet other mail goes through fine, right? Is it only mail for cancun which is weird, or is this a general phenomenon? -- Mark Towfigh, Racal Interlan, Inc. towfiq@interlan.Interlan.COM W: (508) 263-9929 H: (617) 488-2818 uunet!interlan!towfiq "The Earth is but One Country, and Mankind its Citizens" -- Baha'u'llah