[Varnish] #629: Fix the (Open)Solaris event port acceptor
Varnish
varnish-bugs at projects.linpro.no
Sat Jan 23 14:02:38 CET 2010
#629: Fix the (Open)Solaris event port acceptor
----------------------+-----------------------------------------------------
Reporter: slink | Owner: phk
Type: defect | Status: new
Priority: normal | Milestone:
Component: varnishd | Version: 2.0
Severity: normal | Resolution:
Keywords: |
----------------------+-----------------------------------------------------
Comment (by slink):
I've thought about the port_getn timeout again: For nevents == 1 the only
purpose is to wake up again to check for session timeouts, so I've made
the timeout dynamic. I have done tests with max_ts set to 60 and all went
smoothly.
Note that if one wants to increase efficiency by batching up more requests
with (nevents > 1), min_ts and max_ts actually get more important as they
then define not only the tolerance for session expiry, but also a range of
additional latency for processing of keepalive requests.
In short, if you always want best latency, stick to nevents == 1, if you
set nevets > 1, i'd recommend something like 5ms / 20ms for min/max (if
20ms is the max additional latency you want to experience for a client).
{{{
#!C
const struct timespec min_ts = {0L, 5L /*ms*/ * 1000L /*us*/
* 1000L /*ns*/};
const double min_t = 0.005;/* 5 ms*/
const struct timespec max_ts = {0L, 20L /*ms*/ * 1000L /*us*/
* 1000L /*ns*/};
const double max_t = 0.02;/* 20 ms*/
}}}
Note: The new version depends on #630 for TIM_t2ts()
--
Ticket URL: <http://varnish.projects.linpro.no/ticket/629#comment:1>
Varnish <http://varnish.projects.linpro.no/>
The Varnish HTTP Accelerator
More information about the varnish-bugs
mailing list