File unlinking problem

[Ray Gardener]

  ----- Original Message ----- 
  From: Desai, Jason 
  To: MAILSCANNER at JISCMAIL.AC.UK 
  Sent: Friday, February 22, 2002 5:10 PM
  Subject: Re: File unlinking problem


  Was one of the options that you changed the Run As User / Group?  Perhaps the file in the queue was owned by another user?  


  Do you still have the file, so you can check it?
  the file is owned by exim and in the group exim which are the same as my Run As User / Group

  Jase


  Regards

  Ray Gardener

    -----Original Message-----
    From: Ray Gardener [mailto:r.a.gardener at SHU.AC.UK]
    Sent: Friday, February 22, 2002 11:39 AM
    To: MAILSCANNER at JISCMAIL.AC.UK
    Subject: [MAILSCANNER] File unlinking problem


    I had a unusual problem today - unfortunately I am only able to give a description rather that a full log on the instance because the symptoms aren't logged in syslog. (I am logging mail.debug level  BTW)

    Briefly Mailscanner died and wouldn't restart even after several attempts to run check_mailscanner. I will try to paraphase the message it reported but I didn't record it as I was rather flustered at the time. 

    It couldn't resolve an unknown dependency concerning -t option at line 420  of sendmail.pl;  

    that line reads

    unlink "$OutQueue/$file"; 

    and concerns "Removing Duplicates from the outgoing queue:\n";

    after inspecting I decided to look at the outgoing queue; there was a file present which co-incided with the time mailscanner stopped. I moved this to another place and restarted mailscanner with check_mailscanner and mailscanner started to work again.

    1) I think that the time it stop coincided with me killing mailscanner with kill -9 in order to then restart it with different options. Clearly the queue was confused by a duplicate file existing in the outgoing and incoming queues but also clearly the code tries to cater for this situation and in my case failed. Anyone help me with this might have been?

    2) My other point relates to that fact that this error doesn't seem seem to be logged in syslog even though a diagnostic appeared on my console. Should I expect this to be logged somewhere?

    3) is killing with kill -9 advisable?

    4) Finally my queues once restart processed 3.5 thousand messages in a 1.30 hours. I suspect that this could have been faster if I adjusted the number of messages that could be process in a batch from 100 to some number. Is there a sane upper limit for this?

    Technical details

    system: Sunblade 1000
    OS: Solaris 8
    MTA: Exim 3.35
    Version of mailscanner: 3.11-1
    deliver in background set to yes 
    delivery type is queue

    Regards

    Ray Gardener
    Sheffield hallam University
    CIS
    0114 224 4926

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.mailscanner.info/pipermail/mailscanner/attachments/20020222/e986ab34/attachment.html