--------
In message <[email protected]>, Nils Goroll writes:
>Here's the consensus from the bugwash on irc:
>
>- we'll add stats counters for SESS_CLOSE
>- then we'll check on larger production systems which impact increasing
>timeout_req to 7 seconds has by
>  - comparing the RX_TIMEOUT to sess_conn
>  - comparing tcp counters

Related to this, I'm looking at this general area of code in relation
to PROXY and H2, and maybe we should simplify our timers differently.

Right now we have two timeouts:

timeout_idle - close idle connections after this long. [5 sec]

timeout_req - max time to collect req.hdrs (first->last byte) [2 sec].

In reality this is a pretty weird distinction, I'm not even sure how/why
we ended up with it.


Maybe we should instead have two different timeouts:

timeout_first - Time to receive first complete req.hdrs after accept.

timeout_idle - Time to receive subsequent complete req.hdrs after response.

Or maybe even just one:

timeout_idle - Time to receive complete req.hdrs

Comments ?

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
[email protected]         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.

_______________________________________________
varnish-dev mailing list
[email protected]
https://www.varnish-cache.org/lists/mailman/listinfo/varnish-dev

Reply via email to