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

RE: Am thinking of using subversion as a general purpose filing s ystem

From: <andy.glew_at_amd.com>
Date: 2004-04-02 21:30:28 CEST

> Are you indexing file _contents_ or file _names_? Is this
> one of those stupid M$loth programs to improve marginal
> searching by exercising your HD's and CPU at
> all times? I turn those off as a matter of course and do the
> same for Linux indexers; the number of times I have to
> search fulltext is so small compared to
> the amount of CPU cycles it would waste to preindex makes it
> ludicrous.

In my case, it is the UNIX indexers.

I don't turn them off, but typically run at night.
My time is worth more than CPU cycles.
(Hell, man, I *design* CPUs for a living:
I *want* CPU cycles to be cheap.)

> But in either of these cases, the mere accessing the BDB files at
> the filesystem level has no possible way of corrupting the file.

That's what I figured.

 
> > Similarly, sometimes it's more convenient to run backup
> > utilities remotely.
>
> Just like [most?] other databases, you can backup the
> database file itself, but in order to ensure that the data
> is recoverable, you have to follow whatever
> methodology the database supports. Both Oracle and M$SQL
> typically have plugins
> to the backup software or you use an export file. BDB offers
> incremental
> backups in a similar fashion. In most cases, you just
> exclude the database
> files themselves from the backup, since they are not useful
> as backups.

Good. This just puts me back into the usual place with database
backups: you have to create a new backup methodology for them.

Q0: can I continue using my regular filesystem backup
methodology, trusting it will not corrupt the database?
   A: yes: but the database files backed up may be useless.
      The running database will still be good.
Q1: can I tune the backup software to avoid having to
    make useless backups of the database files?
   A: yes
Q2: so then create a new backup methodology for the database.
   :-(
Q3: can the database methodology be seamlessly integrated
   with the system filesystem backup methodology,
   so that you can, e.g. restore to a fixed point in
   time, and just start running the database?
      Or do you have to remember to do a few more steps
   after filesystem repair, to repair each of the dozen
   or so different databases you may have on your backups?
      (And, of course, you have to version control those steps.)

===

Anyway, this conversation is winding down.

Do not use Berkeley DB accesses from different clients of
a network filesystem.

I still have not heard a single answer as to whether it
is okay to use Berkeley DB access from a single client
of a network filesystem, but I suspect it is.

---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Fri Apr 2 21:31:11 2004

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

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