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

Re: libsvn_repos logging -- rethunk

From: Michael Sinz <michael.sinz_at_gmail.com>
Date: 2005-07-19 04:30:39 CEST

On 7/17/05, Ben Collins-Sussman <sussman@collab.net> wrote:
> On Jul 17, 2005, at 9:02 PM, Christopher Ness wrote:
> >
> > Allowing a client to write their own entries in the servers log (on
> > remote machine) seems like a design mistake to me. Couldn't this
> > become
> > a DOS attack as _anyone_ can write a client to fire log messages at a
> > repository server.
> >
> > I might have misinterpreted the comment though,
> Yes you have. We're only talking about clients accessing the
> repository directly via file:/// urls. Users who can do this are
> already free to open the database files in emacs and tweak them
> however they want. There's no "permissions" other than whatever the
> OS file permissions allow.

Ahh, that is good - but still somewhat annoying to play all of these
games with multiple threads/processes writing to the same file.

> svnserve and 'svn subcommand file:///' already share 80% of their
> code -- it's all in the libsvn_repos library. So that's why I'm
> trying to tie logging to libsvn_repos. The following processes all
> open repositories directly for reading/writing:
> mod_dav_svn
> svnserve
> svnlook
> svnadmin
> svn subcommand file:///
> And thus all of these processes should be able to log things when
> they use the repository.
> I think you're assuming that we're talking about clients sending
> "please log this" requests over a network. We're not. :-)

Ok - I am coming late into this whole discussion as I thought it
would not get so bogged down.

I may be a bit "old school" here but why is there so much work
and effort going into re-inventing "syslog" logging. There are
mechanisms within syslog to define the service/facility/etc and
to even filter what gets written to disk, console, remote log server,
etc. Yes, this leaves a small gap (big?) for those systems that
do not have syslog, but then Windows does have application
event logging support, albeit not quite as nice as modern syslog
setups, it is still a "system" solution.

Given a clean definition of the facilities, log levels, and identifiers,
this would provide a nice way to get the logs into a standard location
(configured via syslog.conf) and rotation rules. Plus, people like
me would be much happier since the logs would be a different
machine from the server, thus providing both an extra level of
security and better isolation of responsibilities.

Anyway, just my two cents - I am sure that the Windows world is not
as straight forward as syslog, but it can not be that far behind in
features - syslog is older than even Windows NT 3.1...

Michael Sinz               Technology and Engineering Director/Consultant
"Starting Startups"                          mailto:Michael.Sinz@sinz.org
My place on the web                      http://www.sinz.org/Michael.Sinz
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jul 19 04:31:49 2005

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.