[squid-dev] [RFC] dns_wait_for_all

[Alex Rousskov]

On 09/15/2016 03:50 AM, Amos Jeffries wrote:
> On 15/09/2016 5:11 p.m., Alex Rousskov wrote:
>> On 09/14/2016 07:26 PM, Amos Jeffries wrote:
>>> On 15/09/2016 8:15 a.m., Alex Rousskov wrote:
>>>> Any better ideas or objections to adding dns_wait_for_all?
>>
>>
>>> In principle okay. However, I was intending to redesign the object we
>>> store DNS RR results in to achieve this is a more useful way. 
>>
>> If you are talking about Comm::ConnectionList/serverDestinations, then
>> that object will have to be redesigned to support this proposal, of
>> course. The object would also have to be safely shared across FwdState,
>> peer selection code, and DNS code for the proposed feature to work.
> 
> I mean Dns::LookupDetails, ipcache_addrs and rfc1035_rr.

> Something like:
> 
> * ipcache_addrs::in_addrs becomes a list of rfc1035_rr instead of array
> if IP address. (or an equivalent type storing the same details in easier
> to access type than 'rdata' uses).
> 
> * ipcache_addrs::bad_mask moved into rfc1035_rr for a mask on the
> records in each RR separately.
> 
> * ipcache_addrs merged into Dns::LookupDetails.

That sounds mostly unrelated to the problem at hand. If we need to alter
Dns::LookupDetails, ipcache_addrs, or rfc1035_rr in some significant
way, then we will come back to this discussion. Otherwise, it stays as a
separate/independent not-yet-discussed project.


> The serverDestinations not changing (yet).

I am pretty sure we have to change that field to implement
dns_wait_for_all functionality correctly. For example:

* we need to make its updates (from different jobs!) safe, with
easy-to-trace debugging.

* we need to support an EOF-like semantics to signal whether more
destinations may still be provided.


Thank you,

Alex.