[squid-users] Scaling concurrent TCP sessions beyond ephemeral port range

[Amos Jeffries]

On 9/09/22 11:41, Praveen Ponakanti wrote:
> Hi Alex,
> 
> 
> Thanks for all the help from the squid dev group with upstreaming the 
> enhancement to scale up outbound TCP sessions on Linux with the 
> IP_BIND_ADDRESS_NO_PORT sockopt flag. Our canary instances have been 
> doing great the last few weeks with the code patch prior to merge.
> 
> 
> A few followup questions (not urgent) :
> 
>   * Do we know which 5.x version will include the patch? I do not see it
>     listed in the change log for squid-5.7.

Squid-5 is in "stable" release cycle already which means the changes 
applied to it are quite restricted.

IMO this change is more of a performance optimization than a bug fix, so 
this is being left for Squid-6 which is supposed to start releasing in a 
few months (Feb 2023).



>   * We have a large number of workers (30) to help with handling a
>     high RPS. However, TCP session reuse does not seem to be optimal
>     even with server_persistent_connections enabled as a new outbound
>     session would have to be opened up if the request is proxied by a
>     kid worker that doesn’t already have a connection to that
>     destination. Is there something that can be done to improve this
>     with later versions of squid? Would be glad to help out if anyone
>     has some suggestions.

It sounds to me like your situation is one that this system architecture 
was designed to service: 
<https://wiki.squid-cache.org/ConfigExamples/ExtremeCarpFrontend>



Cheers
Amos