It would be nice if one could run Lighttpd instead of or along side Apache. This would allow for some clients (who might be running Rails apps) to FastCGI built into Lighttpd instead of the module for Apache 2

Hey Kaveh,

This is on the agenda. We have at least one large client of Virtualmin that prefers Lighttpd, and every once in a while I hear rumblings wishing for support...so it is definitely on the agenda. It won't be in the immediate future (we're shooting for stability, bug fixes, and a few additional features, plus platform support in the immediate future), but I'd be surprised if three or four months down the road there isn't some initial support for it.

I also would like to see lighttpd support as a replacement for Apache, make it an install option.

Config options:

Install with Apache only
Install with lighttpd only
Install with Apache on port 80, and lighttpd on 8080
Install with lighttpd on port 80, and Apache on 8080
Install Apache with proxy to lighttpd
Install lighttpd with proxy to Apache


Just my thoughts.

Hey Matthew,

I'll chat with Jamie about how difficult those extra options would be. It's certainly not impossible, but probably would require quite a bit of additional GUI elements to represent all of these options (and we're trying very hard to avoid adding more GUI elements...people are already scared to death by all of the options, and we're actively working to hide, remove and simplify the existing choices). Features and flexibility are the enemy of easy in a lot of cases, and we're getting quite a bit of pushback on the lack of easy in some of our forms. ;-)

So, it's not out of the question, but it won't be in the very near future.

To chime in re: request for Lighttpd support.

Given issues with Apache (memory bloat), that u've been helping me thru Joe (Ta!), i'm very interested in testing Lighttpd.
The only reason for not going with Lighttpd at the start of my project was the 'time tasted' stability of Apache.
And the only reason for not testing now is lack of support in VirtualMin (i just found out).

I'd also suggest that the addition of Lighttpd wld also make VirtualMin a viable CP of choice for the RoR kids, hosts, etc; a very large and growing market.

Hey guys,

Apache isn't really very heavy for what it is and what it does...it just does a lot. It can also be reduced in size, if high load is not expected. You'd be giving up quite a lot of functionality going to Lighttpd. If you aren't using any of that extra functionality, then it's not a problem--but it's often hard to get just the right intersect of "features-we-need" with an "only-the-features-we-need" product.

As for RoR...When mod_fcgid becomes the default execution model after install later this week, RoR will drop right in. RoR runs just fine under Apache+mod_fcgid (I actually developed a RoR application last year that runs that way). The addition of mod_fcgid might actually allow us to shrink Apache a bit, as we could entirely remove mod_php. mod_php is a pretty big component...so stripping it out entirely certainly won't hurt--looks like it takes out about 90MB when removing mod_php5 on my desktop machine.

But, as I mentioned, we're not gonna rule out Light. It's just such a large undertaking, and it would take us away from the other pretty major projects underway: AJAX theme--first release is about a week away!--but we'll still have a month or two of effort required to hook up lots of little niceties, like AJAXed pagination, auto-complete in Usermin webmail address field, real-time updates of graphs and database grids, etc.

Not to mention all of the remaining backend work, and OS support work. FreeBSD users are beginning to sharpen their pitchforks, and Dovecot 1.0 needs to roll out to everyone ASAP so we can switch to 750 permissions for all domain homes. All around, a lot of big projects under way already. I have good reason to believe we'll be making a couple of hires in two or three months, so at that point we may be able to expand our efforts to include another webserver. But I definitely can't commit to Light support anytime in the next couple of months. (And we'll want to see a lot more users who actually want it before adding one more confusing set of features to an already complex system.)