Holger Parplies wrote at about 23:59:49 +0200 on Monday, August 8, 2011:
 > Hi,
 > 
 > Jeffrey J. Kosowsky wrote on 2011-08-08 16:28:28 -0400 [Re: [BackupPC-users] 
 > Too many links : where is the problem ?]:
 > > Holger Parplies wrote at about 19:34:27 +0200 on Monday, August 8, 2011:
 > >  > Hi,
 > >  > 
 > >  > as Jeffrey said, we'll need meaningful information to give meaningful 
 > > answers.
 > > 
 > > Oh my goodness, did Holger Top Quote????? say it isn't so :P
 > 
 > well, I suppose it is *possible* to define top posting that way, but does
 > anyone? I didn't reply to anything and quote the question (or the whole mail)
 > afterwards. I was *trying* to acknowledge that you had given a more
 > comprehensive reply and I was only going to reflect on a minor aspect.
 > 
I was just teasing (hence the :P emoticon)...

 > > FYI, On 32-bit Fedora 12/Linux 2.6.32:
 > >    Ext2/3: MAX=32000
 > >    Ext4: MAX=65000
 > > 
 > > This presumably should be true more generally for any relatively
 > > non-ancient & unhacked version of Linux...
 > 
 > Well, yes, probably, so you only have to figure out whether you have a
 > non-ancient and unhacked version of Linux (or just leave HardLinkMax at the
 > default and not worry about it). In general, I agree that you shouldn't
 > determine things by experiment that can be figured out by reading the
 > documentation. However, almost every Linux distribution seems to add their
 > own patches to the vanilla Linux kernel, and for *this* matter, the 
 > particular
 > limit whichever component of your system may be imposing is quite easy to
 > determine by experiment.

I actually used your perl code to "verify" the source code. :)

I was pleasantly surprised at how fast it ran relative to file creation
but that is understandable. Interestingly, it seemed to run
significantly faster on ext4 than ext3 though deleting the directory
was faster on ext3 than ext4 -- this was not a scientific experiment
though.

That being said, I hope you would agree that the default 32000 number
seems reasonable given that ext2/ext3 is pretty common and seems to be
the least 'max' number of any commonly used hard-link-allowing
filesystem. Given that exceeding the max is typically an infrequent
special case and given that BackupPC seamlessly handles 'overflows' by
creating a second pool instance, there doesn't seem to be any good
reason for a user to change this number unless you (a) know what you
are doing (b) have a special case where the savings from avoiding
creating a second pool chain instance will be substantial...

Finally, out of curiosity, I grepped the BackupPC code base for the
error language "too many links" cited verbatimu by the OP and found that such
a phrase only occurs in the comments and hence is not even a valid
error code... so while this thread has been interesting regarding the
general nature of max fs links, the OP really hasn't given us anything
to help him address his specific problem -- a point that we have both made
originally!

------------------------------------------------------------------------------
uberSVN's rich system and user administration capabilities and model 
configuration take the hassle out of deploying and managing Subversion and 
the tools developers use with it. Learn more about uberSVN and get a free 
download at:  http://p.sf.net/sfu/wandisco-dev2dev
_______________________________________________
BackupPC-users mailing list
BackupPC-users AT lists.sourceforge DOT net
List:    https://lists.sourceforge.net/lists/listinfo/backuppc-users
Wiki:    http://backuppc.wiki.sourceforge.net
Project: http://backuppc.sourceforge.net/