Page 2 of 2 FirstFirst 12
Results 11 to 16 of 16

Thread: postfix/postqueue: fatal: Queue report unavailable - mail system is down

  1. #11
    Join Date
    May 2010
    Posts
    8
    Rep Power
    5

    Default

    Please can you answer a few questions?

    First, is this server a virtual server or a physical server?

    Second, can you post the contents of /etc/fstab showing the mount point(s) for all Zimbra directories?

    Third, is the server's storage for the Zimbra directories on local disk or on a network file system?
    Hi Mark,
    here answers to your questions:
    1) The server is a physical server;
    2) Attached you can find my fstab file. Zimbra folder is inside "dati" folder, connected with opt directory via a simple symlink.
    3) the server's storage for Zimbra directories is on a local raid array

    I don't know if is usefull for you, but:
    - Reading system log file,i find out that the problem was originated by a loss of power due to the crash of the ups.
    - I tried, during my research of a solution, to change permission to the pid folder inside /opt/data/postfix/spoll. At the moment that i deleted the master.pid file, the permissions were 0777.
    - When I were trying to start postfix manually, no output was generated, it simply didn't start (it seems that the postfix process finded the pid file and realized that it was already running without performing more checks).

    If you need more information, ask without problems.

    Giorgio
    Attached Files Attached Files

  2. #12
    Join Date
    Sep 2006
    Location
    477 Congress Street | Portland, ME 04101
    Posts
    1,374
    Rep Power
    11

    Default

    We stay away from ReiserFS but are currently testing a Postfix installation where all drives are now mounted with the "noatime" option on ext3.

    (If available, Postfix reads the atime (access time) information to do things like enabling the flush daemon to slowly feed deferred emails back into the system. Mutt reportedly uses atime if available but not much else does these days.)

    In this one case, we are seeing that the file system time starts to differ from the system time by several minutes or several hours. Postfix gets confused by the difference and crashes. This happens every few weeks and the logs have timestamps both way ahead and far behind each other. BTW, when no atimes are available, Postfix stays running.

    In our test system it is not clear how the file system and system time stamps diverge, but since atime is not required for Postfix to run, we generally disable it as doing so also improves performance.

    Since you are not using a time server on your system and your mount options in /etc/fstab enable atime, it is possible that divergent file system and system times contributed to your issue. I'd therefore suggest adding the noatime mount option to your two drives and see if the divergent time stamps in the logs no longer appear.

    Hope that helps, and please let us know if you have any more crashes,
    Mark

  3. #13
    Join Date
    May 2010
    Posts
    8
    Rep Power
    5

    Default

    Thank you for your exellent answer: I didn't know anything about this problem with atime attribute.
    I have a personal love for ReiserFS, but I know that ReiserFS offers much more troubles than ext3.
    I, now, have only few questions more (I'm interested in the problem because I manage 6 different Zimbra servers):
    why postfix restarted after I deleted master.pid? The problem is caused by a wrong atime notation on a file (fsck should fix it I suppose, but not in my case) or by an error in the postfix execution in this conditions?

  4. #14
    Join Date
    Sep 2006
    Location
    477 Congress Street | Portland, ME 04101
    Posts
    1,374
    Rep Power
    11

    Default

    Quote Originally Posted by misterg73 View Post
    Thank you for your exellent answer: I didn't know anything about this problem with atime attribute.
    I have a personal love for ReiserFS, but I know that ReiserFS offers much more troubles than ext3.
    I, now, have only few questions more (I'm interested in the problem because I manage 6 different Zimbra servers):
    why postfix restarted after I deleted master.pid? The problem is caused by a wrong atime notation on a file (fsck should fix it I suppose, but not in my case) or by an error in the postfix execution in this conditions?
    I agree ReiserFS had a lot of promise, but when the main author gets sent to prison for murdering his wife that puts a stop to active development. Further, we found that ReiserFS did not always recover well from crashes, so we stayed with ext3.

    I believe Postfix restarting after you deleted the master.pid is normal. Personally, I wouldn't do that on a Zimbra system. There are various Zimbra commands that can start and stop portions of Zimbra independently, and without having to restart all of Zimbra, e.g: zmmtactl and zmmailboxdctl.

    Zimbra is really best thought of as an appliance, controlled only using Zimbra-supplied tools. Sure, there are lots of familiar components (Postfix, OpenLDAP, MySQL, Amavis, ClamAV, SpamAssassin, etc.), but even something as simple as adding a line to Postfix's main.cf file requires either an ability to that through the Admin Console or, if not available, a good knowledge of Zimbra's inner architecture.

    Hope that helps,
    Mark

  5. #15
    Join Date
    May 2010
    Posts
    8
    Rep Power
    5

    Default

    There are various Zimbra commands that can start and stop portions of Zimbra independently, and without having to restart all of Zimbra, e.g: zmmtactl and zmmailboxdctl.
    I know that Zimbra tools are the best way to controll and manage Zimbra server.I know that fixing perm with postfix tool rather than zmfixperms causes many problem, etc.. But belive me, I tried everyting I know to restart the server, included zmmtactl. There is an advanced manual about Zimbra, something more specilaized than Administator's guide? In this guide, this time, I wasn't able to find a solution even if the guide untill now had been very usefull...

  6. #16
    Join Date
    Oct 2013
    Posts
    4
    Rep Power
    2

    Default

    Quote Originally Posted by misterg73 View Post
    I think I have find a solution:

    /opt/zimbra/data/postfix/spool/pid

    even if postfix wasn't running.
    I deleted the file and everyting restart.
    today I have the same problem - mta status on web portal say "running" but postfix does not. I read your solution with the pid file and investigate that on my system

    del pid file
    restart zimbra
    = all ok

    I want to say: thank you!

    -js

Similar Threads

  1. Replies: 7
    Last Post: 11-21-2011, 12:21 AM
  2. Replies: 7
    Last Post: 02-03-2011, 06:01 AM
  3. Replies: 8
    Last Post: 12-01-2010, 08:14 PM
  4. Replies: 20
    Last Post: 03-18-2008, 05:37 AM
  5. Mail not getting to mail boxes
    By ehults in forum Installation
    Replies: 5
    Last Post: 10-31-2005, 08:24 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
  •