[squid-users] Squid 4 pconn_lifetime questions
Alex Rousskov
rousskov at measurement-factory.com
Tue May 14 14:07:58 UTC 2019
On 5/14/19 12:25 AM, johnr wrote:
> If in the context of this directive became idle means "done processing the
> previous request" then how is the pconn_lifetime directive different than
> the client_idle_pconn_timeout and server_idle_pconn_timeout (other than
> affecting both at the same time)?
The former directive limits the lifetime of a connection. The latter
directives limit the time a connection can stay in the idle pool.
In other words, the clock for the first directive starts ticking when
the connection is created while the clock for the other two directives
starts ticking when the connection is placed in the idle connection pool.
> If my question doesn't make sense, then
> perhaps what I'm asking is better stated as: how do the pconn_lifetime and
> client_idle_pconn_timeout interact?
There should be virtually no interaction: The former limit is checked
just when a connection becomes idle and Squid decides whether to pool
the connection or close it. The latter timeout is checked for the
already pooled connections.
GitHub pull requests clarifying documentation are welcomed.
Alex.
More information about the squid-users
mailing list