Actually, I was hoping for a clue from Henrik on how the Squid process gets access to a file outside of the chroot during a reconfigure. I believed Henrik when he said it needed to be outside of the chroot. Just looking to understand Squid's chroot implementation a more deeply. ;-)

Rick G. Kilgore wrote:
Works just as if it was not chroot jailed at all. Actually with the conf file in the chroot I had more problems accessing external ACL and other oddities.



Joe Cooper wrote:

Henrik Nordstrom wrote:

On Tue, 31 Aug 2004, Joe Cooper wrote:

resolve.conf) that Squid relies on (it could be that shared libraries are pulled in before Squid chroots, and so they might not be needed--Henrik wrote the chroot code I think, or at least maintains it now, maybe he'll chime in with clarification).




If you use the chroot directive in squid.conf then only logs, cache and a dev/null node is minimally required within the chroot directory structure. It is also a good idea to set up a syslog socket within the chroot (man syslogd).

The squid configuration file and any data referenced from there should be outside of the chroot directory, and unless you use any helpers no libraries is required either.



Out of curiosity: without squid.conf in the chroot, how does a -k reconfigure work?







Reply via email to