Re: Pinger disable patch - 2nd try (where is my mind ???, forgot the patch....)

The netdb code uses more than ICMP to build it's database. It also
periodically collects netdb information with the peers, and also uses ICP
to collect up to date RTT information (always enabled in the queries)

I don't know how well netdb will operate without ICMP support, but I guess
it should work fine for child/sibling caches.

Some of your patch is currently redundant.. and I am a little bit worried
about how netdbPingDomain() will react.

icmpRecv()

should NOT be patched. Only called if there is a ICMP socket opened and
packets have been received on it..

icmpPing()
  
never seems to be used anywhere..?? But it does not hurt to patch it..

icmpSourcePing()

is already protected in the netdb code, but I guess it does not hurt to
protect it again.. perhaps a debug message should be added if any of
these are called with no pinger enabled?

further, ALLOW_SOURCE_PING cannot be enabled without manually adding it to
the DEFINED in the Makefile..

On Sun, 21 Jul 2002, Guido Serassio wrote:

> Hi,
>
> This patch add the option to disable pinger when Squid is compiled with
> ICMP enabled.
> I have tested Squid without pinger, and seems to run fine, but I'm not
> absolutely sure that it can't be side effect running netdb code without
> pinger, Your opinion ?
Received on Sun Jul 21 2002 - 06:05:35 MDT