From: Pedro AF. Cunha (Pedro.Cunha@decsis.pt)
Date: Wed Sep 03 2003 - 06:38:41 EDT
Hi all,
Once again many thanks for your cooperation.
The solution was to properly configure the /etc/services and /etc/clua_services (thanks to Tom Smith) and adding the following entries:
/etc/services
smtps 465/tcp
submission 587/tcp
/etc/clua_services
smtps 465/tcp in_multi # sendmail secure smtp
submission 587/tcp in_multi,static # sendmail message submission port
This solved the problem.
All the replies follow below.
Special thanks to:
Martin Rønde Andersen [martin.roende@mail.dk]
Dr Thomas.Blinn@HP.com [tpb@doctor.zk3.dec.com]
Fred N. van Kempen [Fred.van.Kempen@microwalt.nl]
Tom Smith [smith@cag.lkg.dec.com]
-------------------------------------------------------------------------------------------------------------------------
Hello Pedro.
sendmail 8.12 opens additional listener ports beyond the port 25 opened by the Tru-64 standard sendmail 8.9. These additional ports are not specified in the standard Tru64 /etc/clua_services as "in_multi", which is required to be able to open the same port number on multiple members. What is happening is that the first instance of sendmail to start grabs those ports cluster-wide, peventing any other sendmail instances from opening them and starting.
Add the follwing to /etc/clua_services:
smtps 465/tcp in_multi # sendmail secure smtp
submission 587/tcp in_multi,static # sendmail message submission port
submission 587/udp in_multi,static # sendmail message submission port
The UDP port is actually unnecessary, I think.
You will probably also want to add these ports to
/etc/services or to your NIS services map.
-Tom
-------------------------------------------------------------------------------------------------------------------------
Hmm... yes, if the second node tries to activate the daemon, on the same virtual cluster IP address, then it cannot use the same port number. Either Sendmail has to "wait" for the port to become available (read: the other node died), or it has to bind to the port number on the node local IP address...
--fred
-------------------------------------------------------------------------------------------------------------------------
As I recall, some special work was done on sendmail to make it cluster aware (for the version that ships in the standard product). I suspect you can't just take the "freeware" sendmail and get it to work in your cluster, without reverse engineering what the TruCluster product does to make sendmail work in a cluster.
I have NO personal knowledge of the details, but as I recall, the last time I added a member to a cluster, when I booted the member for the first time, there were a number of messages displayed about setting up sendmail to be be cluster aware or capable, I don't have a log of the messages nor do I know just which component (probably a script) sent them out or what it was doing at the time.
Tom
Dr. Thomas P. Blinn + Tru64 UNIX Software + Hewlett-Packard Company
Internet: tpb@zk3.dec.com, thomas.blinn@compaq.com, thomas.blinn@hp.com
110 Spit Brook Road, MS ZKO3-2/W17 Nashua, New Hampshire 03062-2698
Alpha Hardware Platforms and I/O Phone: (603) 884-0646
ACM Member: tpblinn@acm.org PC@Home: tom@felines.mv.net
Worry kills more people than work because more people worry than work.
Keep your stick on the ice. -- Steve Smith ("Red Green")
My favorite palindrome is: Satan, oscillate my metallic sonatas.
-- Phil Agre, pagre@alpha.oac.ucla.edu
Yesterday it worked / Today it is not working / UNIX is like that
-- apologies to Margaret Segall
Opinions expressed herein are my own, and do not necessarily represent
those of my employer or anyone else, living or dead, real or imagined.
------------------------------------------------------------------------------------------------------------------------
This might be in the cluster alias area.
Check with /etc/clua_services for the portnumber, and see if you have latest patch kit. Over time there has been numeous cluster alias problems .
M. v. h./ regards - Martin Rønde Andersen
The Unix era continues ... http://unixera.com
-------------------------------------------------------------------------------------------------------------------------
Hi again,
After a little further investigation on this I've come up with this:
- The second trucluster member starting the sendmail daemon shows the following message in the mail.log file:
Aug 28 18:38:26 member2 sm-mta[525933]: daemon MSA: problem creating SMTP socket Aug 28 18:38:31 member2 sm-mta[525933]: NOQUEUE: SYSERR(root): opendaemonsocket: daemon MSA: cannot bind: Address already in use
...and then the daemon dies.
Does any one have seen this problem? How can I turn around this?
Many thanks,
Pedro Cunha.
-------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------
Original Question
Hi all,
We are running Trucluster V5.1B, patch kit 2, and sendmail 8.12.8 and at the second time that a connection to the smtp port is made the sendmail daemon dies on the cluster member that is not answering to the request.
for example:
member 1
529512 ?? S 0:00.02 sendmail: accept -L sm-mta -bd -q25m
529524 ?? I 0:00.00 sendmail: Queue -L sm-msp-queue -Ac -q 25m
member 2
529512 ?? S 0:00.02 sendmail: accept -L sm-mta -bd -q25m
529524 ?? I 0:00.00 sendmail: Queue -L sm-msp-queue -Ac -q 25m
now doing two consecutive telnet to cluster alias on port 25 (member 1 is answering), and after this,
member 1
529512 ?? S 0:00.02 sendmail: accept -L sm-mta -bd -q25m
529524 ?? I 0:00.00 sendmail: Queue -L sm-msp-queue -Ac -q 25m
member 2
1051777 ?? I 0:00.00 sendmail: Queue -L sm-msp-queue -Ac -q 25m
Any ideas of what is hapening???
Many thanks in advance,
Pedro Cunha.
Pedro Cunha
(Systems Engineer)
Decsis - Sistemas de Informação, Lda.
Rua das artes gráficas, 162
4100-091 Porto
Portugal
Tel.: +351 226076850
Fax: +351 226076859
This archive was generated by hypermail 2.1.7 : Sat Apr 12 2008 - 10:49:34 EDT