Re: relayd TCP reconnect?

Re: relayd TCP reconnect?

[Jérémie Galarneau]

On Wed, Mar 25, 2015 at 11:16 AM, Jesper Derehag <jderehag@hotmail.com> wrote:
> Hi All,
>
> I know the issue with relayd reconnects have been discussed earlier here:
> http://lists.lttng.org/pipermail/lttng-dev/2014-October/023650.html
>
> But has there been any further development on this?
> I would actually go as far as calling it a bug and submitting it to redmine tracker, but want to get approval here first.
>

Hi Jesper,

There has been no work done on this yet.

> I would argue that its *always* a good idea to do reconnects due to that you really dont have any end-to-end control of the TCP flow.
> Some random router might decide to spuriously drop your connection, not to mention more modern queuing disciplines which might randomly throw away packets.
>
> On automated systems, that would mean that you have to somehow automagically detect that you got a disconnect, and then externally restart the session, which seems like a very weird workaround...
>
> I would be happy to file such a issue to the tracker if people thinks its a good idea, (and depending on if I have the time I *might* look into solving it myself).
> But it might be some time away in that case..

Sure, feel free to open a ticket. We'll definitively have to look into
recovering gracefully at some point.

Jérémie

>
> Regards,
> Jesper
>
>
>
> _______________________________________________
> lttng-dev mailing list
> lttng-dev@lists.lttng.org
> http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev



-- 
Jérémie Galarneau
EfficiOS Inc.
http://www.efficios.com

_______________________________________________
lttng-dev mailing list
lttng-dev@lists.lttng.org
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

FW: relayd TCP reconnect?

[Jesper Derehag]

Forwarding to lttng-dev as well..

----------------------------------------
> From: jderehag@hotmail.com
> To: jeremie.galarneau@efficios.com
> Subject: RE: [lttng-dev] relayd TCP reconnect?
> Date: Thu, 26 Mar 2015 08:20:52 +0000
>
> Great, added new RM issue:
> https://bugs.lttng.org/issues/883
>
> Thanks!
> /Jesper
>
> ----------------------------------------
>> Date: Wed, 25 Mar 2015 14:14:27 -0400
>> Subject: Re: [lttng-dev] relayd TCP reconnect?
>> From: jeremie.galarneau@efficios.com
>> To: jderehag@hotmail.com
>> CC: lttng-dev@lists.lttng.org
>>
>> On Wed, Mar 25, 2015 at 11:16 AM, Jesper Derehag <jderehag@hotmail.com> wrote:
>>> Hi All,
>>>
>>> I know the issue with relayd reconnects have been discussed earlier here:
>>> http://lists.lttng.org/pipermail/lttng-dev/2014-October/023650.html
>>>
>>> But has there been any further development on this?
>>> I would actually go as far as calling it a bug and submitting it to redmine tracker, but want to get approval here first.
>>>
>>
>> Hi Jesper,
>>
>> There has been no work done on this yet.
>>
>>> I would argue that its *always* a good idea to do reconnects due to that you really dont have any end-to-end control of the TCP flow.
>>> Some random router might decide to spuriously drop your connection, not to mention more modern queuing disciplines which might randomly throw away packets.
>>>
>>> On automated systems, that would mean that you have to somehow automagically detect that you got a disconnect, and then externally restart the session, which seems like a very weird workaround...
>>>
>>> I would be happy to file such a issue to the tracker if people thinks its a good idea, (and depending on if I have the time I *might* look into solving it myself).
>>> But it might be some time away in that case..
>>
>> Sure, feel free to open a ticket. We'll definitively have to look into
>> recovering gracefully at some point.
>>
>> Jérémie
>>
>>>
>>> Regards,
>>> Jesper
>>>
>>>
>>>
>>> _______________________________________________
>>> lttng-dev mailing list
>>> lttng-dev@lists.lttng.org
>>> http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
>>
>>
>>
>> --
>> Jérémie Galarneau
>> EfficiOS Inc.
>> http://www.efficios.com
>
 		 	   		  

relayd TCP reconnect?

[Jesper Derehag]

Hi All,

I know the issue with relayd reconnects have been discussed earlier here:
http://lists.lttng.org/pipermail/lttng-dev/2014-October/023650.html

But has there been any further development on this? 
I would actually go as far as calling it a bug and submitting it to redmine tracker, but want to get approval here first.

I would argue that its *always* a good idea to do reconnects due to that you really dont have any end-to-end control of the TCP flow. 
Some random router might decide to spuriously drop your connection, not to mention more modern queuing disciplines which might randomly throw away packets.

On automated systems, that would mean that you have to somehow automagically detect that you got a disconnect, and then externally restart the session, which seems like a very weird workaround...

I would be happy to file such a issue to the tracker if people thinks its a good idea, (and depending on if I have the time I *might* look into solving it myself). 
But it might be some time away in that case..

Regards,
Jesper