Re: WCCPv2 issues
ons 2006-07-26 klockan 11:27 +1200 skrev Reuben Farrelly: > > Regarding the mask based assignments, wonder how complex mask arrays a > > typical equipment supports. In the specs is an unbounded array. > > Does this help? > > http://www.cisco.com/en/US/products/hw/switches/ps4324/products_configuration_guide_chapter09186a008062cfc6.html > > especially > > "WCCP version 2 standard allows for support of up to 256 distinct masks. > However, a Catalyst 4500 series switch only supports mask assignment table > with > a single mask." I suppose this means a single mask/value set. Wonder if there is any limit on the size of the value set.. hopefully not too small or load distribution will be a bit troublesome. A single mask is sufficient provided the value set can be sufficiently large. But in worst case this means a single mask value per cache member server, which kind of limits possible setups to power of 2 cache servers as all other configurations will have IP areas which can't be redirected if only a single mask value per cache server is supported. A quick math. They say it supports up to 32 cache servers. Then hopefully they support at least 64 mask values to allow for some kind of distribution. Btw, not sure where they found the supposed standard limit of 256.. nothing mentioned in the last draft I have access to, and the field with is 32 bits... The only 256 limit mentioned is the hash size for hash distribution which is using 256 buckets (fixed size). Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel
Re: WCCPv2 issues
tis 2006-07-25 klockan 23:09 +0200 skrev Henrik Nordstrom: > Turns out the router capability processing has never been tested > See also > http://www.squid-cache.org/mail-archive/squid-users/200607/0450.html as > there was some questionmarks while decoding the WCCP2_HERE_I_AM sent by > Squid. > > May be other bugs as well in these WCCP2 exchanges. I don't quite get > some of the IOS error messages, but maybe it's just incomplete data. I think all pieces of the WCCP2 issue has now fallen into place, except the questionmarks in the message referenced above. Awaiting testing of the proposed patch before commit. From what it looks the equipment he uses simply not compatible with Squid at the moment as it seems it only supports mask based assignment, and we only support hash based assignment. Regarding the mask based assignments. I wonder how complex mask arrays a typical equipment supports. In the specs is an unbounded array only limited by the UDP packet size. The mask specifications can build very powerful expressions, but also harder to distribute among the farm members. Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel
WCCPv2 issues
Bug #1696 could use a couple of other eyes. Turns out the router capability processing has never been tested (return instead of break in a switch statement.. see the bug), and something seems fishy there (squid complaining on what to me looks like correct packets).. See also http://www.squid-cache.org/mail-archive/squid-users/200607/0450.html as there was some questionmarks while decoding the WCCP2_HERE_I_AM sent by Squid. May be other bugs as well in these WCCP2 exchanges. I don't quite get some of the IOS error messages, but maybe it's just incomplete data. Regards Henrik signature.asc Description: Detta är en digitalt signerad meddelandedel