https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6268
--- Comment #1 from Justin Mason <[email protected]> 2010-01-05 07:29:45 UTC --- (In reply to comment #0) > I'm unsure if the change to a die for the lack of a user_dir was intentional. > So anyone know if/why/etc? it was not, I think. here's the specific error message: Dec 30 09:35:24 mail1 mimedefang-multiplexor[4448]: Slave 0 stderr: config: path "/root/.spamassassin" is inaccessible: Permission denied Dec 30 09:35:24 mail1 mimedefang-multiplexor[4448]: Slave 0 stderr: config: error accessing /root/.spamassassin: Permission denied Dec 30 09:35:24 mail1 mimedefang-multiplexor[4448]: Slave 0 died prematurely -- check your filter rules Dec 30 09:35:24 mail1 mimedefang[9301]: Error from multiplexor: ERR No response from slave Dec 30 09:35:24 mail1 mimedefang-multiplexor[4448]: Reap: slave 0 (pid 4449) exited normally with status 13 (SLAVE DIED UNEXPECTEDLY) Dec 30 09:35:24 mail1 mimedefang-multiplexor[4448]: Slave 0 resource usage: req=1, scans=1, user=0.316, sys=0.012, nswap=0, majflt=0, minflt=7034, maxrss=0, bi=0, bo=0 ( http://www.mail-archive.com/[email protected]/msg13454.html ) see line 1847 in http://svn.apache.org/viewvc/spamassassin/trunk/lib/Mail/SpamAssassin.pm?annotate=894117 for the change details: 1839 felicity 410643 if (!$self->{dont_copy_prefs}) { 1840 dbg("config: using \"$fname\" for user state dir"); 1841 felicity 410676 } 1842 felicity 410643 1843 mmartinec 697056 my $stat_errn = stat($fname) ? 0 : 0+$!; 1844 if ($stat_errn == 0 && !-d _) { 1845 die "config: $fname exists but is not a directory\n"; 1846 } elsif ($stat_errn != 0 && $stat_errn != ENOENT) { 1847 die "config: error accessing $fname: $!\n"; 1848 } else { # does not exist, create it 1849 felicity 410676 # not being able to create the *dir* is not worth a warning at all times 1850 mmartinec 568336 eval { 1851 mkpath($fname, 0, 0700); 1; 1852 } or do { 1853 my $eval_stat = $@ ne '' ? $@ : "errno=$!"; chomp $eval_stat; 1854 dbg("config: mkdir $fname failed: $eval_stat"); 1855 }; 1856 jmason 3805 } so it appears to have been Mark's toughening-up of error conditions change, r697056: http://svn.apache.org/viewvc?view=revision&revision=697056 http://svn.apache.org/viewvc/spamassassin/trunk/lib/Mail/SpamAssassin.pm?r1=697055&r2=697056&; Revision 697056 - (view) (download) (annotate) - [select for diffs] Modified Fri Sep 19 11:26:46 2008 UTC (15 months, 2 weeks ago) by mmartinec File length: 65312 byte(s) Diff to previous 690090 (colored) [Bug 5981] Improved error detection and reporting That's https://issues.apache.org/SpamAssassin/show_bug.cgi?id=5981 . I think the appropriate fix here would be to change that die() to a dbg(); as the following comment notes, failure to create the dir there is not even worth a warning at all times, let alone a die(). We handle write failures appropriately later, when saves of the journal files/databases are attempted. I'll attach a patch. -- Configure bugmail: https://issues.apache.org/SpamAssassin/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are the assignee for the bug.
