Results 1 to 5 of 5

Thread: failure to accept a message with "Illegal address syntax" in log

  1. #1
    grisha is offline New Member
    Join Date
    Jan 2012
    Posts
    4
    Rep Power
    3

    Default failure to accept a message with "Illegal address syntax" in log

    Hi,

    I'm migrating the e-mail server from aging Sun Enterprise /Solaris-8/sendmail to newer SunFire/CentOS/ZCS-7.1.3_GA.

    The new system works quite well and stable during last 3 months, with exception of only one, but very important, to [not] be ignored, problem.

    I seam to run into a weird issue; I can clearly understand its nature, but I never experienced it with Sendmail based servers - I have a large number of 3..5 year old Sun servers in our clusters, whose ILOM (management/monitoring) system sends alerts as "ilom-alert@192.168.0.15" (address scrambled) thus resulting in getting this message in the log file:

    Jan 23 22:10:20 myZimbraName postfix/smtpd[10195]: connect from sun01.scrambled.com[192.168.0.15]
    Jan 23 22:10:20 myZimbraName postfix/smtpd[10195]: warning: Illegal address syntax from sun01.scrambled.com[192.168.0.15] in MAIL command: <ilom-alert@192.168.0.15>
    Jan 23 22:10:20 myZimbraName postfix/smtpd[10195]: disconnect from sun01.scrambled.com[192.168.0.15]

    Under Solaris/sendmail - I've never experienced the above problem; moreover - I've tried sending ILOM alerts from one of those servers to one of my lab/test servers' newly configured "sendmail" - still no problems.

    Under Zimbra - I'm pulling my hair off during last 3 months trying all possible configuration "ideas" and hints to make ZCS allow those "invalid" addresses (anything in format "<from-name@IP-a.d.d.res>") - nothing helps.

    Have anybody succeeded doing so? - I mean, allowing Zimbra to accept messages from "broken" hardware-based e-mail originators?

    As a very, very stupid way around I've already tried my "backup" plan - configure all those servers' ILOM to send all alerts to some "lightly" loaded virtual server running CentOS / Sendmail whose only task was to accept the alert for delivery and then re-forward it to my main (ZCS) server (I've used "procmail" mechanism). Stupid, but works... it involves an additional and unwanted point of failure and delay though.

    I'm totally open to reinstall, reconfigure or try anything less stupid that may allow receiving those ILOM alerts directly to my ZCS.

    Thanks!

  2. #2
    phoenix is online now Zimbra Consultant & Moderator
    Join Date
    Sep 2005
    Location
    Vannes, France
    Posts
    23,566
    Rep Power
    57

    Default

    Quote Originally Posted by grisha View Post
    I seam to run into a weird issue; I can clearly understand its nature, but I never experienced it with Sendmail based servers - I have a large number of 3..5 year old Sun servers in our clusters, whose ILOM (management/monitoring) system sends alerts as "ilom-alert@192.168.0.15" (address scrambled) thus resulting in getting this message in the log file:
    The Zimbra server is RFC compliant and the email address you're using is an Illegal address and get's rejected for that reason. You need to modify the sending application to use a compliant address.
    Regards


    Bill


    Acompli: A new adventure for Co-Founder KevinH.

  3. #3
    grisha is offline New Member
    Join Date
    Jan 2012
    Posts
    4
    Rep Power
    3

    Default failure to accept a message with "Illegal address syntax" in log

    Hello,

    Thanks for replying to my message.

    I'm aware that Zimbra is RFC compliant and that the client's address is encoded incorrectly.

    But unfortunately there is absolutely NO way "to modify the sending application to use a compliant address" because the "sending application" is a hardware card with a dumb monitoring software (ILOM) installed by Sun in millions of their servers between 2004 through 2009 while there is no more Sun as a company anymore - it's Oracle who doesn't care updating older hardware's firmware.

    Our support contracts for more than 120 of those servers expired after 2008/2009 and there was no incentive to continue those contracts over because there were expensive and inefficient.

    My question was whether I could modify any setting within Zimbra's configuration to get around this issue - as I wrote earlier, the "sendmail" based server perfectly works in this situation, but it would be very inconvenient and add another point of failure if I configured ILOM cards of all those servers to send their alerts to another 'sendmail" server first - just to be re-forwarded to my main Zimbra setup.

    I don't mind purchasing a commercial Zimbra license if that required to perform the above asked modifications - I understand that the "postfix" is an extremely flexible software while I simply don't know which setting to change (during last 20 years I've been using 'sendmail" based servers).

    Is there any way of doing so? (while browsing this forum I saw that there were other requests of doing such a configuration, but I couldn't find any replies to those requests).

    Thank you.

    Regards,
    Gregory.

  4. #4
    resolve_numeric_domain is offline Starter Member
    Join Date
    Jun 2012
    Posts
    1
    Rep Power
    3

    Default Try resolve_numeric_domain

    I had this problem as well setting up SMTP alerting from a Sun X4540's ILOM processor forwarding to an SMTP relay running Postfix 2.3.3 on Centos 5.6.

    Fortunately the resolve_numeric_domain setting described in Postfix Configuration Parameters setting fixed everything. I just add to /etc/postfix/main.cf

    # attempt to fix Sun ILOM
    resolve_numeric_domain = yes

    and then did a

    # /usr/sbin/postfix reload

    and my X4540's ILOM was able to send SMTP alerts as desired.

  5. #5
    grisha is offline New Member
    Join Date
    Jan 2012
    Posts
    4
    Rep Power
    3

    Default

    Quote Originally Posted by resolve_numeric_domain View Post
    I had this problem as well setting up SMTP alerting from a Sun X4540's ILOM processor forwarding to an SMTP relay running Postfix 2.3.3 on Centos 5.6.

    Fortunately the resolve_numeric_domain setting described in Postfix Configuration Parameters setting fixed everything. I just add to /etc/postfix/main.cf

    # attempt to fix Sun ILOM
    resolve_numeric_domain = yes

    and then did a

    # /usr/sbin/postfix reload

    and my X4540's ILOM was able to send SMTP alerts as desired.
    Wow! Amazing, I'd never ever be able to figure that out on my own. Probably - no one would.
    Problem solved!

    Thank you very much.

Thread Information

Users Browsing this Thread

There are currently 1 users browsing this thread. (0 members and 1 guests)

Similar Threads

  1. Temporary message delivery failure
    By goetzi in forum Administrators
    Replies: 40
    Last Post: 06-25-2007, 10:41 AM
  2. Recover single message from redo log?
    By shanson in forum Administrators
    Replies: 1
    Last Post: 06-06-2006, 08:07 AM
  3. Replies: 20
    Last Post: 05-07-2006, 11:34 PM
  4. Is it started or not
    By kwelipatton in forum Installation
    Replies: 10
    Last Post: 03-28-2006, 11:11 PM
  5. system failure: getDirectContext
    By avisser in forum Installation
    Replies: 3
    Last Post: 10-12-2005, 05:32 AM

Tags for this Thread

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •