Hi all, new to Nginx here and am investigating whether it can possibly
replace our current configuration in production.
I’ve searched a bit on this but haven’t come up with much:
We run a web cluster that serves files (updates) for download for our
software that clients run, and the files are updated every 30-60
seconds. Our cache-control headers for certain file types are set to 5
seconds to make sure clients are always pulling the most up-to-date
update.
Right now we use Apache on the front end and Squid behind Apache on the
same system (this setup duplicated on many systems). This allows every
incoming http request to Apache to look to Squid for the file, and squid
either serves from cache or pulls from the source creating the files if
the cached file is expired.
Our requirement is that due to the 5 second cache expiration settings,
we need squid to continue to serve stale files if a cached file expires
and the backend server isn’t responding (server down, etc) so that our
clients servers don’t log http errors. Squid permits this via the
stale-if-error directive.
Our current solution works, but wastes system resources since this means
that for every connection to a web server, a corresponding 2nd
connection is created (to squid). What we’d prefer to do is replace the
Apache+Squid solution with Nginx, but I haven’t found anything in the
documentation or forums/mailing lists with any mention of if Nginx can
be configured to serve stale cached content if the source is down.
Can anyone shed any light? Thanks!
Posted at Nginx Forum: