Hi Baptiste. Baptiste have written on Wed, 24 May 2017 11:34:31 +0200:
> Hi all, > > Over the last few weeks, I entirely reworked the internal resolver of > HAProxy to make it more flexible. > The main driver for this is to add more features related to DNS > use-cases (SRV records, scale in / scale out a backend, DNS > converter, etc...) and also to make it more efficient internaly and > more friendly with DNS servers ;) > > With this in mind, I performed the following changes: > - DNS tasks are now autonomous, they are not triggered by the health > check any more, this means we can enable DNS resolution without > enabling health monitoring on a server > - DNS responses are now kept in memory, when they are not an error... > so many servers using the same hostname resolution will use a > response in the cache if it is fresh enough > - full "anonymisation" of the requesters. Up to now, DNS code was > though for servers only (well layers were pretty well defined). I > added an abstraction layer between the resolution and the requester, > so the requester could be of any HAProxy internal type (backend, > bind, ...). > - management of requester through queues: 2 queues are available: the > run queue and the wait queue. If a requester needs a resolution, he > will register himself to the run queue, and if he is the first one, > then the resolution is triggered > - a pool of resolution is now linked to a resolvers section. Memory is > allocated at configuration parsing time. If the pool is too small for > your configuration, an error is returned. > > I still have a few minor things to work on, but since HAProxy Tech > guys needs my code to move forward on their contribution, I'm > publishing it right now. Wow great work. I have take a look into the code and have just some questions about calloc in [PATCH 03/11] & [PATCH 07/11] In the function dns_alloc_resolution is calloc used, would the use of haproxy pools bring any benefit? [PATCH 08/11] +static int dns_cache_size = 1024; /* arbitrary DNS cache size */ For the future maybe there could be a glolbal.tune.dns-cache-size or something similar. Do I have understand the code right that you have per resolver a cache and this resolver belongs to a backend/frontend. So in case several backends have the same server names all resolver requests this server individual. There is no 'global dns cache' for all or I missed something. > Please give it a try and report any issues you may spot :) > > Baptiste Regards Aleks
