On 7/11/2024 12:15 PM, Mathew Heard wrote:
Do you happen to know if there remains any gap in obvious capability
provided by a module like jdomain compared to this?
An obvious difference is that the jdomain directive performs resolution
on demand instead of periodic polling, but that's a design choice rather
than a feature gap. Other that that I don't see anything missing.
Perhaps it would be worth checking to ensure nothing obvious is not
implemented?
The one that I see is the ability to control if a resolution is IPv4,
IPv6 or mixed. Is this something that would be useful for this feature?
That is already implemented in the resolver directive. If the http block
scope is too large, one of the patches in the series allows configuring
resolver per upstream.
upstream backend {
zone upstream_dynamic 64k;
resolver 127.0.0.1 ipv4=off ipv6=on;
server example.com resolve;
};
On Fri, 12 Jul 2024 at 02:40, Aleksei Bavshin <a.bavs...@nginx.com
<mailto:a.bavs...@nginx.com>> wrote:
On 7/9/2024 9:22 AM, Roman Arutyunyan wrote:
> Hi,
>
> On Mon, Jul 08, 2024 at 06:20:58PM +0400, Roman Arutyunyan wrote:
>> Hi,
>>
>> On Thu, Jun 13, 2024 at 03:28:56PM -0700, Aleksei Bavshin wrote:
>>> # HG changeset patch
>>> # User Ruslan Ermilov <r...@nginx.com <mailto:r...@nginx.com>>
>>> # Date 1392462754 -14400
>>> # Sat Feb 15 15:12:34 2014 +0400
>>> # Node ID 56aeae9355df8a2ee07e21b65b6869747dd9ee45
>>> # Parent 02e9411009b987f408214ab4a8b6b6093f843bcd
>>> Upstream: re-resolvable servers.
>>>
>>> Specifying the upstream server by a hostname together with the
>>> "resolve" parameter will make the hostname to be periodically
>>> resolved, and upstream servers added/removed as necessary.
>>>
>>> This requires a "resolver" at the "http" configuration block.
>>>
>>> The "resolver_timeout" parameter also affects when the failed
>>> DNS requests will be attempted again. Responses with NXDOMAIN
>>> will be attempted again in 10 seconds.
>>>
>>> Upstream has a configuration generation number that is
incremented each
>>> time servers are added/removed to the primary/backup list.
This number
>>> is remembered by the peer.init method, and if peer.get detects
a change
>>> in configuration, it returns NGX_BUSY.
>>>
>>> Each server has a reference counter. It is incremented by
peer.get and
>>> decremented by peer.free. When a server is removed, it is
removed from
>>> the list of servers and is marked as "zombie". The memory
allocated by
>>> a zombie peer is freed only when its reference count becomes zero.
>>>
>>> Re-resolvable servers utilize timers that also hold a reference. A
>>> reference is also held while upstream keepalive caches an idle
>>> connection.
>>>
>>> Co-authored-by: Roman Arutyunyan <a...@nginx.com
<mailto:a...@nginx.com>>
>>> Co-authored-by: Sergey Kandaurov <pluk...@nginx.com
<mailto:pluk...@nginx.com>>
>>> Co-authored-by: Vladimir Homutov <v...@nginx.com
<mailto:v...@nginx.com>>
>
> [..]
>
>>> diff --git a/src/http/ngx_http_upstream_round_robin.h
b/src/http/ngx_http_upstream_round_robin.h
>>> --- a/src/http/ngx_http_upstream_round_robin.h
>>> +++ b/src/http/ngx_http_upstream_round_robin.h
>>> @@ -14,8 +14,23 @@
>>> #include <ngx_http.h>
>>>
>>>
>>> +typedef struct ngx_http_upstream_rr_peers_s
ngx_http_upstream_rr_peers_t;
>>> typedef struct ngx_http_upstream_rr_peer_s
ngx_http_upstream_rr_peer_t;
>>>
>>> +
>>> +#if (NGX_HTTP_UPSTREAM_ZONE)
>>> +
>>> +typedef struct {
>>> + ngx_event_t event; /* must be
first */
>>> + ngx_uint_t worker;
>
> Missed this last time. This field should be removed since all
resolving is in
> worker #0.
Unfortunately, that would break the ABI compatibility between OSS and
Plus. Replacing the field with yet another NGX_COMPAT_BEGIN isn't any
better than leaving it in the opensource code.
>
>>> + ngx_str_t name;
>>> + ngx_http_upstream_rr_peers_t *peers;
>>> + ngx_http_upstream_rr_peer_t *peer;
>>> +} ngx_http_upstream_host_t;
>>> +
>>> +#endif
>>> +
>>> +
>>> struct ngx_http_upstream_rr_peer_s {
>>> struct sockaddr *sockaddr;
>>> socklen_t socklen;
>>> @@ -46,7 +61,12 @@ struct ngx_http_upstream_rr_peer_s {
>>> #endif
>>>
>>> #if (NGX_HTTP_UPSTREAM_ZONE)
>>> + unsigned zombie:1;
>>
>> I suggest declaring this as in other similar places:
>>
>> ngx_uint_t zombie; /* unsigned
zombie:1; */
>>
>>> +
>>> ngx_atomic_t lock;
>>> + ngx_uint_t id;
>>
>> This field is not used in open source nginx and should not be
added or assigned.
>>
>>> + ngx_uint_t refs;
>>> + ngx_http_upstream_host_t *host;
>>> #endif
>>>
>>> ngx_http_upstream_rr_peer_t *next;
>
> [..]
>
> --
> Roman Arutyunyan
> _______________________________________________
> nginx-devel mailing list
> nginx-devel@nginx.org <mailto:nginx-devel@nginx.org>
> https://mailman.nginx.org/mailman/listinfo/nginx-devel
<https://mailman.nginx.org/mailman/listinfo/nginx-devel>
_______________________________________________
nginx-devel mailing list
nginx-devel@nginx.org <mailto:nginx-devel@nginx.org>
https://mailman.nginx.org/mailman/listinfo/nginx-devel
<https://mailman.nginx.org/mailman/listinfo/nginx-devel>
_______________________________________________
nginx-devel mailing list
nginx-devel@nginx.org
https://mailman.nginx.org/mailman/listinfo/nginx-devel
_______________________________________________
nginx-devel mailing list
nginx-devel@nginx.org
https://mailman.nginx.org/mailman/listinfo/nginx-devel
