[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: Anti-FUD: SVN ate my repository

From: Toby Johnson <toby_at_etjohnson.us>
Date: 2004-09-16 15:54:12 CEST

Jani Averbach wrote:

>I wrote a document to tackle down a FUD that "after some time SVN
>repositories will get corrupted".
>If you could read and commend it that would be nice.
If I knew nothing about Subversion, I would feel *less* comfortable with
it after reading your document. Since your stated goal is "anti-FUD",
I'm assuming that isn't what you were trying to accomplish.

I could basically paraphrase the top section as "sometimes when people
think their repository is corrupted, it's really just wedged, but
sometimes it really is corrupted and you've lost data". Hardly a very
comforting way to start out a FAQish document.

You then go on to describe wedged repositories and why it might happen.
But then you go into "real" corruption, where you give a good point
about NFS/SMB shares and disk space, but the rest is basically saying
"yeah, sometimes data is lost, and I really don't know why".

If I'm looking into Subversion and I see "Unclassified corrupted
repositories: I think that I have seen a couple cases when there has not
been any particular reason for database corruption", I'm running the
other way as fast as I can. What are these instances of corruption you
"think you've seen"? From there, you talk about Berkeley DB being "too
brittle". Again, not very reassuring.

I wouldn't say any of this is factually incorrect, but it's all
presented in the wrong way. If you're trying to dispel FUD, then make no
mention of any of this until the very end and say something like "of
course, no computer is 100% bulletproof so be sure to always make
backups of your data", which you've already described somewhat. Saying
"yeah, I think I've seen some data corruption but it might have been a
burnt CPU or something" is not a good way to dispel myth.

Unless you know of any particular cases of corruption, then don't
speculate on it; that will only make others more wary. I'm not
suggesting you hide the truth, but I know of no true cases of corruption
that were Subversion or BerkeleyDB's fault.

To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Sep 16 15:54:33 2004

This is an archived mail posted to the Subversion Dev mailing list.

This site is subject to the Apache Privacy Policy and the Apache Public Forum Archive Policy.