Thanks Eric.
I saw that snippet yesterday, but for whatever reason my mind did not absorb the information in a useful manner. A day later, some sleep, and time away from the problem helps me see the text in a different light. That is, my solution is embedded in that text snippet, but yesterday was not intuitively obvious to me.
When I copied the cache from NT4 to Slackware, I also copied the
swap.state file. In my defense I had no idea the file would be incompatible with the version of squid I installed on Slackware. The snippet above does provide a clue. The solution seems to be 1) copy the cache, 2) do
not copy the
swap.state file, 3) start squid manually (
squid -NCd1) to send output directly to the screen and provide immediate feedback , and 4)
wait---as long as necessary for squid to rebuild the cache index (
swap.state).
I have a cache that is almost 1 GB in size, so I'll be sure to run the rebuilding process while I am doing something else. However, I am reasonably certain this approach will work. Thanks again!
Question:
I'm confused by the squid.conf directive
cache_dir.
At my NT4 box, currently I have squid configured with:
cache_dir ufs o:/squid/cache 1024 16 256
My original understanding is that the
1024 (megabytes) refers to the
total squid cache size. That is, when squid fills the sublevels to the point of being within 95% of 1024 MB (1 GB) of cached files, squid then will begin the process of deleting least recently used files. But yesterday I ran across a snippet that implied the
1024 number refers to each sublevel. That is, of the 16 sublevels defined above, squid will store 1024 MB of files in each sublevel. Which is correct?
The embedded instructions in squid.conf does not explain this in detail, or I am not reading the text correctly. My reading remains the former interpretation, but if I am wrong and the
1024 refers to each sublevel, then I need to know because I don't have the drive space to store
1024 MB x
16 (16 GB) worth of files.
Question:
Unlike SquidNT, yesterday I quickly learned that installing squid in GNU/Linux requires me to pay attention to file and directory permissions. After some web surfing, I noticed that some people recommend creating a new user and group account called
squid, and assigning the log and cache directories permissions to that user and group. Some people just leave things be and use
nobody/nogroup. Anybody here have any insight in the difference and a preference? I understand why squid refuses to run as root. I'm just asking which method people prefer: 1) leave things as is with
nobody/nogroup or create
squid/squid and modify the squid.conf
cache_effective_user and
cache_effective_group directives?
