Zitat von "W.C.A. Wijngaards" <[email protected]>:
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On 11/11/2010 09:41 AM, [email protected] wrote:
Zitat von "W.C.A. Wijngaards" <[email protected]>:
No, it only uses the 8m to set a parameter on the socket itself (that
overrides the kernel variable).
It does override the default, but must be lower then the *_max setting, no?
Normally, yes, on Linux unbound supports the _FORCE variation, which can
override the *_max (if you are root, therefore it sets this option
before dropping privileges).
I could implement such an option so-sndbuf; would it be easier to just
make it be equal to so-rcvbuf? (saves another config entry?)
Not sure. The receive sockets are low in number and tend to higher
latency if the machine is under load and could not process the
incoming packets fast enough. The send sockets could be much higher in
number (outgoing-range?) and should in case of UDP only fall behind if
the available bandwidth is saturated.
So i guess the send buffer per socket used should be smaller than the
receive buffers...
Regards
Andreas
_______________________________________________
Unbound-users mailing list
[email protected]
http://unbound.nlnetlabs.nl/mailman/listinfo/unbound-users
