On Wed, 26 Jan 2011 18:27:56 -0800 (PST), Scott Lehman wrote: > Hi, > > I need to control bandwidth for many logical groups of hosts based on > time, but want to change the available bandwidth while a download is in > progress (which can take an hour or more). > > My initial plan was to: > -use external ACLs to tag requests from each logical host group > -use time-based ACLs to select the appropriate delay pool > > The only catch is that the ACLs only appear to be evaluated at request > processing time (reasonable). I'm basically hoping re-evaluate which pool > a long-running request belongs in, or change the pool parameters over time. > > Is something like this possible with Squid? > > My searches haven't turned up much so far (dynamic delay pool patch looked > different), and my only idea is to restart Squid with a different config > and hope clients can resume where they left off. With delay pools that is pretty much the only solution not involving code changes to Squid. If you want to do such changes a regular event every N seconds (configurable) that re-evaluates ll active client delay pools would seem to be useful. QoS alternatives to delay pools exist in all the current releases of Squid. tcp_outgoing_tos can set a TOS marking for system traffic control to use based on ACL. The timing and grouping may be more possible at the system packet handling level. Amos
