Varnish 3.0.1 Backend conditional requests
geoff at uplex.de
Mon Jan 9 13:23:27 CET 2012
-----BEGIN PGP SIGNED MESSAGE-----
On 01/ 9/12 11:55 AM, Jocelyn De La Rosa wrote:
> Hi,I have also tested the experimental-ims branch and it's working great!
> Are there any (hidden) bugs that prevent its use in production mode?
> We are very interested by this feature, so if some work is needed we could
> do it.
Actually, there may be some hidden bugs. Kristian Lyngstol set up some
automated stress tests for the various branches a few months ago, and
the experimental-ims branch was not passing all of them. (The branch has
been out of the automation for a while, I presume because there haven't
been any updates.)
Also the branch is not currently up to date with the changes that have
been done on the master branch, especially the changes in memory
These matters are on my to do list. %^)
Let's please get in touch offline if you plan to make contributions.
phk recently announced plans on the dev list for an upcoming new
release, including adding the IMS feature. These issues will need to be
resolved by then, of course.
IMS actually almost made it into 3.0, but there was some unresolved
discussion then about whether the configuration in VCL as currently
implemented is the best way to go. There was an idea under consideration
that, rather than have stale_obj and the two different keep and grace
intervals after TTL elapses, there could be a new vcl sub (say,
vcl_stale()) to cover both the grace and keep cases.
I've had some feedback from people who've tried the experimental-ims
branch that they're happy with the configuration the way it is. At any
rate, a consensus about that will of course be necessary.
** * * UPLEX - Nils Goroll Systemoptimierung
Tel +49 40 2880 5731
Mob +49 176 636 90917
Fax +49 40 42949753
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (SunOS)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org/
-----END PGP SIGNATURE-----
More information about the varnish-misc