Hi Rune,
I don't get any further with this without more input.
First a question: were *all* bindings from 1.1.2 missing after the reboot, or
only the first 6 ones (the ones in the "bulk" publication of message #12646 (in
the big pcap file)?).
If the latter is the case, then we know that it is
They were not in there after the reboot, might not have been there before
either.
Only way to actually get it working was to restart whichever application has
the missing registration on 1.1.2.
-Original Message-
From: Jon Maloy [mailto:jon.ma...@ericsson.com]
Sent: Monday, April 04, 2
They might not have been.
-Original Message-
From: Jon Maloy [mailto:jon.ma...@ericsson.com]
Sent: Monday, April 04, 2016 11:44 AM
To: Rune Torgersen; 'Jon Maloy'; tipc-discussion@lists.sourceforge.net
Cc: erik.hu...@gmail.com; Richard Alpe; Parthasarathy Bhuvaragan
Subject: RE: [tipc-di
Thank you Rune,
I think my theory was wrong. I can now see that the dropped items actually were
withdrawals, not publications, that were sent out just before the 1.1.2
rebooted, of course because the server application was being killed at that
moment.
They were probably queued because the corres
Until now, the requests sent to topology server are queued
to a workqueue by the generic server framework.
These messages are processed by worker threads and trigger the
registered callbacks.
To reduce latency on uniprocessor systems, explicit rescheduling
is performed using cond_resched() after MA
> -Original Message-
> From: Rune Torgersen [mailto:ru...@innovsys.com]
> Sent: Monday, 04 April, 2016 09:53
> To: Jon Maloy; 'Jon Maloy'; tipc-discussion@lists.sourceforge.net
> Cc: erik.hu...@gmail.com; Richard Alpe; Parthasarathy Bhuvaragan
> Subject: RE: [tipc-discussion] tipc nametab
Tested and reviewed. Nice work Erik! :)
Richard
On 2016-04-02 10:52, Erik Hugne wrote:
> Patch #1 aims to fix a potential issue with deferred updates being pushed
> to the wrong namespace
> Patch #2 should solve the problem with stale updates in the defer queue after
> node down,
>
The capture was all traffic between the two server after reboot.
I still have the full capture (only filtered by mac), how much in front/back do
you want?
-Original Message-
From: Jon Maloy [mailto:jon.ma...@ericsson.com]
Sent: Sunday, April 03, 2016 9:36 PM
To: Jon Maloy; tipc-discussi
The test set up I have are two servers with SuperMicro X10DRL-i motherboards,
each having two Xeon E5-2630V3 8 core CPU's, and 64GB of memory.
I am running Ubuntu 16.04 (beta). Each server also have 10 1G ethernet
interfaces, but only one was active in this case, and I only use one as a
bearer.
Until now, the requests sent to topology server are queued
to an unbound ordered workqueue by the generic server framework.
These messages are processed by worker threads and trigger the
registered callbacks.
To reduce latency on uniprocessor systems, explicit rescheduling
is performed using cond_r
10 matches
Mail list logo
