Hi Willy, William,
Op 2-10-2018 om 3:56 schreef Willy Tarreau:
it's important to cut this into pieces
to figure what it's showing.
A little update on this issue split in 2 pieces.
-Connection and request counters to low when ran as regtest from
varnishtest (bug?)
It turns out that starting haproxy from varnishtest, and using -W
master-worker mode, actually creates 2 processes that are handling
traffic. That explains that a large part of connections isn't seen by
the other haproxy instance and stats showing to low amounts of
connections. Bisecting it seems to fail on this commit: b3f2be3 ,
perhaps @William can you take a look at it? Not really sure when this
occurs in a 'real' environment, it doesn't seem to happen when manually
running haproxy -W, but still its strange that when varnisttest is
calling haproxy this occurs.
-Request counter to high (possibly a improvement request?)
The http_end_txn_clean_session(..) is called which increments the
counter on a finished http response, and i was testing with 2 different
methods for 1.8 and 1.9 due to missing length converter i used in my 1.9
test, which makes the comparison unfair. Sorry i didn't realize this
earlier i thought it did 'more or less' the same, that seems to have
been 'less'. Together with that i found the numbers odd/unexpected i
assumed a bigger problem that it actually seems to be, perhaps its not
even that bad, haproxy is 'preparing' for a second request over the
keep-alive connection if i understand correctly. Which eventually
doesn't happen, but is counted. Maybe that is a point that can be
improved in a future version if time permits.? Or would it even be
expected to behave like that?
Regards,
PiBa-NL (Pieter)
