>
Yes, I would suggest to use an increasing timeout. Starting with zero, up to a maximum. That isn't an uncommon pattern.
Sounds reasonable.
So, thanks for the input. I've created a Hono issue for this:
https://github.com/eclipse/hono/issues/1210
Best regards
Carsten Lohmann
Engineering Cloud Services 4 Bosch IoT Hub (INST/ECS4)
Bosch Software Innovations GmbH | Ullsteinstr. 128 | 12109 Berlin |
GERMANY |
www.bosch-si.com
Sitz: Berlin, Registergericht: Amtsgericht Charlottenburg; HRB 148411 B
Aufsichtsratsvorsitzender: Dr.-Ing. Thorsten Lücke; Geschäftsführung: Dr. Stefan Ferber, Michael Hahn, Dr. Aleksandar Mitrovic
From: hono-dev-bounces@xxxxxxxxxxx <hono-dev-bounces@xxxxxxxxxxx>
On Behalf Of Jens Reimann
Sent: Donnerstag, 25. April 2019 18:05
To: hono developer discussions <hono-dev@xxxxxxxxxxx>
Subject: Re: [hono-dev] Reconnect interval
Yes, I would suggest to use an increasing timeout. Starting with zero, up to a maximum. That isn't an uncommon pattern.
A simple pattern could be:
On connect, reset relay to "min". On connection failure increment the delay if it is below "max".
I'm not actually a developer here, so take this with a grain of salt, but this sounds like something an exponential back-off could address properly. I.e., wait x milliseconds, then 2x, then 4x... etc.
Cheers,
Bor
-----Ursprüngliche Nachricht-----
Von: hono-dev-bounces@xxxxxxxxxxx <hono-dev-bounces@xxxxxxxxxxx> Im Auftrag von Kai Hudalla
Gesendet: Donnerstag, 25. April 2019 17:03
An: hono-dev@xxxxxxxxxxx
Betreff: Re: [hono-dev] Reconnect interval
On Thu, 2019-04-25 at 14:21 +0000, Lohmann Carsten (INST/ECS4) wrote:
> Hi,
>
> I would like to get some opinions on how we should deal with the
> *reconnect
> interval* concerning AMQP connections between Hono components.
>
> Currently, when for example the connection between a protocol adapter
> and the device registry fails or is closed, the re-connect attempt
> happens only after 500ms (that is the DEFAULT_RECONNECT_INTERVAL_MILLIS = 500).
>
> I guess this was introduced for the scenario that the device registry
> is indeed unavailable and for example currently being (re)started, so
> that an immediately-after re-connect attempt would probably fail.
>
> But, there are other scenarios:
> - The device registry could be deployed with multiple instances
> (Kubernetes
> pods) and even if one instance is currently not responding, the next
> connection attempt could reach another, fully available instance.
> - The device registry could have actively closed the connection with a
> "access token expired" response, meaning there is no reason to wait
> for a re-connect attempt.
>
>
sounds reasonable
> IMHO, these two scenarios are quite common. I would therefore suggest
> to get rid of the reconnect interval altogether, always trying to
> re-connect immediately.
What about simply setting it to 0 in our example deployment? People could then still decide whether they want to adjust it according to their needs?
> This would for example prevent the protocol adapters from giving out
> 503 Unavailable responses while waiting for the reconnect interval to
> elapse before connecting again to the device registry.
>
> WDYT?
>
>
> Best regards
>
> Carsten Lohmann
>
> Engineering Cloud Services 4 Bosch IoT Hub (INST/ECS4) Bosch Software
> Innovations GmbH | Ullsteinstr. 128 | 12109 Berlin | GERMANY |
> www.bosch-si.com
>
> Sitz: Berlin, Registergericht: Amtsgericht Charlottenburg; HRB 148411
> B
> Aufsichtsratsvorsitzender: Dr.-Ing. Thorsten Lücke; Geschäftsführung: Dr.
> Stefan Ferber, Michael Hahn, Dr. Aleksandar Mitrovic
>
>
>
> _______________________________________________
> hono-dev mailing list
> hono-dev@xxxxxxxxxxx
> To change your delivery options, retrieve your password, or
> unsubscribe from this list, visit
> https://www.eclipse.org/mailman/listinfo/hono-dev
_______________________________________________
hono-dev mailing list
hono-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/hono-dev
_______________________________________________
hono-dev mailing list
hono-dev@xxxxxxxxxxx
To change your delivery options, retrieve your password, or unsubscribe from this list, visit
https://www.eclipse.org/mailman/listinfo/hono-dev
--
Jens Reimann
Principal Software Engineer / EMEA ENG Middleware
Werner-von-Siemens-Ring 14
85630 Grasbrunn
Germany
phone: +49 89 2050 71286
_____________________________________________________________________________
Red Hat GmbH, www.de.redhat.com,
Registered seat: Grasbrunn, Commercial register: Amtsgericht Muenchen, HRB 153243,
Managing Directors: Paul Argiry, Charles Cachera, Tom Savage, Michael O'Neill
|
