[Dirvish] Explanation Needed of Error Message

[Paul Slootman]

On Thu 09 Aug 2012, Rich Shepard wrote:
> Date: Thu, 9 Aug 2012 07:55:57 -0700 (PDT)
> From: Rich Shepard <rshepard at appl-ecosys.com>
> To: dirvish at dirvish.org
> Subject: [Dirvish] Explanation Needed of Error Message

>    When I logged in this morning I saw messages on the console (prior to my
> logging in) related to /dev/sdb1 but did not write them down. This device is
> my dirvish backup drive (mounted as /media/hd0). /var/log/messages shows
> only:
> 
> Aug  8 00:30:01 salmo kernel: kjournald starting.  Commit interval 5 seconds
> Aug  8 00:30:01 salmo kernel: EXT3-fs (sdb1): using internal journal
> Aug  8 00:30:01 salmo kernel: EXT3-fs (sdb1): mounted filesystem with
> ordered data mode

Those are normal mount messages.

>    However, only two of six partitions were backed up; a message from root in
> my inbox tells me:
> 
> salmo-usr:default:20120808-2200: cannot open log file /media/hd0/salmo-usr/20120808-2200/log

Hmm, it would be helpful if dirvish printed the reason for the failure;
changing line 922 (may vary a bit) would help:

--- dirvish     2012-04-27 12:41:39.538456434 +0200
+++ /tmp/dirvish        2012-08-09 17:27:10.488463212 +0200
@@ -919,7 +919,7 @@
	my ($file, @messages) = @_;
	my $message;
 
-	open (LOGFILE, '>>' . $file) or seppuku 20, "cannot open log file $file";
+	open (LOGFILE, '>>' . $file) or seppuku 20, "cannot open log file $file: $!";
	for $message (@messages)
	{
		print LOGFILE $message, "\n";

Something similar near lines 1126 and 1133 will help show the reason for
config file open failures.

>    The drive is only about 38% full. I've not before encountered an error

A full drive could prevent creating a log file, but not opening an
existing config file.

Perhaps a fsck (forced) may give some explanation.


Paul