Program failure (-6)

[Greg McCann]

On 1/14/2005 at 8:21 PM David Relson <relson at osagesoftware.com> wrote:

>Given the "-6" failure and how often negatives are errno.h values with
>sign reversed, I took a look and found:
>
>  #define  ENXIO  6  /* No such device or address */
>
>which isn't obviously useful :-<  However, you _might_ be running out of
>drive space or your wordlist might be nearing the max file size allowed
>by procmail.  There's info in the FAQ on using postconf to show the
>postfix limits.  What do "df" and "ls -lh wordlist.db" and postconf
>report?

df shows plenty of space available:

# df -m
Filesystem           1M-blocks      Used Available Use% Mounted on
/dev/sda2                  989       636       303  68% /
/dev/sda1                  989        22       917   3% /boot
/dev/sda3                10833      7825      2459  77% /home
none                       251         0       251   0% /dev/shm
/dev/sda5                  989        17       922   2% /tmp
/dev/sda6                 8864      1221      7193  15% /usr
/dev/sda7                10833       308      9975   3% /var

I just rebuilt wordlist.db to see if it would help.

Before:
-rw-rw-rw-    1 root     root     35266560 Jan 14 16:58 wordlist.db

After:
-rw-rw-rw-    1 root     root     15400960 Jan 14 16:59 wordlist.db

>Given that you're using procmail, have you statments like the following in
>procmailrc?
>
>  VERBOSE=yes
>  LOGABSTRACT=yes
>  LOGFILE=/var/log/procmail

I have LOGABSTRACT on but I don't have VERBOSE on at the moment.  If the problem persists, I will try it.

However the problem seems to be rare, intermittent, and not able to be duplicated.  My logs show that it has occurred 72 times in the past 30 days, out of almost 100,000 emails.

I don't know if rebuilding the database helped, but I re-sent myself the same message and it got through.

Any chance this could be some kind of file locking problem?  That might explain the relatively rare occurrence and my inability to duplicate it.

Thank you, David.


Greg