ADSM-L

[ADSM-L] Error aborts backup, instead of missing a file: right, wrong, indifferent?

2007-08-24 12:58:56
Subject: [ADSM-L] Error aborts backup, instead of missing a file: right, wrong, indifferent?
From: "Allen S. Rout" <asr AT UFL DOT EDU>
To: ADSM-L AT VM.MARIST DOT EDU
Date: Fri, 24 Aug 2007 12:57:06 -0400
Howdy.

One of my backup clients started getting an error recently:

08/20/07   23:39:05 TransErrno: Unexpected error from lstat, errno = 79
08/20/07   23:39:06 PrivIncrFileSpace: Received rc=131 from fioGetDirEntries:  
/export/research40  
/compbio2/nfisher/CarneyData/spikeStuff/Rat004/DownSampled200SeizureData
08/20/07   23:39:06 ANS1999E Incremental processing of '/export/research40' 
stopped.

08/20/07   23:39:06 ANS1028S An internal program error occurred.


we've traced this back to some files which had objectively insane
metadata (negative atime (why does anything interpret that as signed,
anyway?), mtimes after 2038, etc).  So we agree that the files in
question have data which it is not possible to correctly capture.


But we're wondering if it's reasonable to halt processing at this
point.  I look at that and interpret the decision as "I'm getting an
insane read off the filesystem, I don't want to back up insanity;
all-stop, and call a human".

On the other hand, it'd be nice for me to be able to say in response
to that alert something like "OK, I'm the human, and I'm saying just
skip over that.  Yes, I know this means I'm exposing myself to
possibly undefined behavior on restore".


Opinions?   Any client developer want to chime in with some clue? :)


- Allen S. Rout

<Prev in Thread] Current Thread [Next in Thread>