On Fri, 23 Nov 2012 10:57:20 +0100, <[email protected]>
wrote:
We're doing some tests with Varnish, and I'm trying to understand some
timings.
Just thinking out loud...
Please correct me where I'm wrong.
If:
1) you are sending a request that hashes to the same object &&
2) your backend is taking a long time to answer &&
3) the object is not in cache when the first request hits
then it could be that varnish is "herding", waiting for the backend
to respond before satisfying all the requests at once.
You should see "Herd(?:ing)" in the varnishlog output.
However:
With a very low traffic (<5 req/sec), I got sometime a delay
of several *seconds* in the fourth number of ReqEnd [1].
The 4th number is time from the session start, so
if your client is using KeepAlive, it could well be that
it connects at t0 and sends n requests where the nth request
starts at t0 + 5s.
Or could be a mix of keepalive + herding?
--
Cosimo
_______________________________________________
varnish-misc mailing list
[email protected]
https://www.varnish-cache.org/lists/mailman/listinfo/varnish-misc
