I think I was unclear in my request for "meanings"...:-)
On Sat, Jul 15, 2006 at 11:50:37PM -0700, Linda W wrote:
I was trying to track down a problem and got distracted on squid status
codes. I was curious on how to interpret these:
~10,000 TCP_MISSes of various sorts
~ 1,500 TCP_CLIENT_REFRESH_MISSes of various sorts
1 TCP_HIT (object found in cache)
1 TCP_NEGATIVE_HIT (lack of object(404) found in cache)
Sums: 15,000 MISSES 2 HITS
Cache stats: 123,005 Entries, 5.6GB (5661932k)
Hit:Request Ratios: All => 1:7500 Positive => 1:15,000
Does this means my cache "accelerates" retrieval, 0.007% of the time?
Isn't this very much on the low side? Is that, even, "normally"
possible, or is it likely something is misconfigured or broken?
By "normally", I visit news sites with many static elements,
including small graphics. Shouldn't, (I would think) those
elements be getting retrieved locally? At "worst", the cache
might check to see if the item is "current" if it had previously
expired, but be able to satisfy the request if it found the item
was unchanged?
In 1500 of the entries, it appears I asked for a refresh, and the
item in the cache was expired _and_ "not current". If a client
is "refreshing", shouldn't the cache be able to satisfy most
of the page's objects? (like graphical elements and such)?
Is it possible (or likely) that most (or, maybe, all(?!)) of the
sites I normally visit ("/.", BBC, GoogleNews, among others)
would be using no-cache (or equivalent) on _all_ of their content.
My refresh pattern is:
refresh_pattern ^ftp: 1440 20% 10080
refresh_pattern ^gopher: 1440 0% 1440
refresh_pattern . 0 20% 4320
Shouldn't that mean that HTML content could (should?) be held
over an hour if it was an "old", static object?
Seems like my squid cache isn't doing me much good as is...;^/
I'm sure it's done better in the past, but wouldn't this be
unusual for a weeks worth of web browsing by the same person?
Thanks,
Linda
