Precisely -- one Textdrive system probably hosts a few dozen user
domains, so they use Apache as a secure and stable front-end, and just
proxy requests to each backend account's own lighttpd instance.
That also allows them to give individual hosting customers privileges
to maintain administer "their" lighttpd instance (to restart, switch
RAILS_ENV settings, etc.), without being able to break other
customers' setups if they screw up.
I don't want to make this a TextDrive flame war... No, actually I do,
but I'll refrain.
However, I want to point out a few things that a new TxD customer
might not know, and might get bit by. A few of us brigadiers have
had these problems; it makes things (kinda) nicer if you know them up
front:
- Don't assume you can run non-production environments. TxD doesn't
want you to do debugging on their servers, so changing RAILS_ENV may
not be okay.
- Expect your long-lived dispatch.fcgi processes to be killed at
random. This is part of their acceptable-use policy, whereby you
can't use more than some percentage of CPU time, but they don't do a
good job telling you much about it. You'll probably find this when
you wonder why your site has gone down. (And don't set up a cronjob
that will watch your lighttpd and restart it; that'll get you bitch-
slapped.)
- There are a *lot* more than "a few dozen user domains" on each TxD
server.
- Once you've set up a domain using webmin, don't go messing with
them via ssh (except for the actual web content). For example, when
you cancel your TxD account, don't delete the directories by hand, or
you'll cause the main Apache config to be busted.
- Read the forums, especially when you're trying debug some
problem. Many important system changes are not documented on TxD's
blog or support pages, and certainly not by email directly to the
customer.
A little grumpy this Halloween night,
--John
_______________________________________________
PDXRuby mailing list
[email protected]
IRC: #pdx.rb on irc.freenode.net
http://lists.pdxruby.org/mailman/listinfo/pdxruby
