From swmike at swm.pp.se  Fri Feb  9 08:45:14 2018
From: swmike at swm.pp.se (Mikael Abrahamsson)
Date: Fri, 9 Feb 2018 17:45:14 +0100 (CET)
Subject: [Netalyzr] how to run single test in the CLI version
Message-ID: <alpine.DEB.2.20.1802091740370.3478@uplift.swm.pp.se>


Hi,

I've successfully used Netalyzr CLI client for a lot of years and it's 
been really useful. Now lately I had a device crash on one of the tests. 
I'd like to isolate exactly what causes the device to crash when it's 
seeing certain traffic through it, so I'd like to run one test at a time 
and make sure I find the minimum set of packets to provide to the vendor 
to show how to replicate the crash.

http://netalyzr.icsi.berkeley.edu/cli.html suggests that there is a "-m 
<mode>" but no description of that this mode is.

My suspicion is that the "-m <mode>" might give me the option of just 
running a single test, but I can't find a syntax that works for me (I 
tried running "strings" on the java files and using grep to find a mode, 
but didn't find anything useful.

Any suggestions on how to make the CLI version run a single test?

Thanks.

-- 
Mikael Abrahamsson    email: swmike at swm.pp.se

From christian at icir.org  Fri Feb  9 10:07:01 2018
From: christian at icir.org (Christian Kreibich)
Date: Fri, 9 Feb 2018 10:07:01 -0800
Subject: [Netalyzr] how to run single test in the CLI version
In-Reply-To: <alpine.DEB.2.20.1802091740370.3478@uplift.swm.pp.se>
References: <alpine.DEB.2.20.1802091740370.3478@uplift.swm.pp.se>
Message-ID: <7d7f8318-86ba-3dfc-56d1-eae7f3514d18@icir.org>

Hi Mikael,

On 02/09/2018 08:45 AM, Mikael Abrahamsson wrote:
> 
> Hi,
> 
> I've successfully used Netalyzr CLI client for a lot of years and it's
> been really useful. 

That's great to hear!

> Now lately I had a device crash on one of the tests.
> I'd like to isolate exactly what causes the device to crash when it's
> seeing certain traffic through it, so I'd like to run one test at a time
> and make sure I find the minimum set of packets to provide to the vendor
> to show how to replicate the crash.
> 
> http://netalyzr.icsi.berkeley.edu/cli.html suggests that there is a "-m
> <mode>" but no description of that this mode is.
> 
> My suspicion is that the "-m <mode>" might give me the option of just
> running a single test, but I can't find a syntax that works for me (I
> tried running "strings" on the java files and using grep to find a mode,
> but didn't find anything useful.

Unfortunately that's not what the mode flag does -- this flag is 
primarily for labeling a session as it enters our database, to allow us 
to identify sets of sessions that are coming in from a certain 
experiment, setting, partner site, etc. It may also mean a slightly 
tweaked set of tests, but there's currently no way to run a single test. 
If you send us some of your session IDs (feel free to keep these 
off-list), we'd be happy to take a look regarding what might be going wrong.

Best,
-C.

From swmike at swm.pp.se  Fri Feb  9 11:27:12 2018
From: swmike at swm.pp.se (Mikael Abrahamsson)
Date: Fri, 9 Feb 2018 20:27:12 +0100 (CET)
Subject: [Netalyzr] how to run single test in the CLI version
In-Reply-To: <7d7f8318-86ba-3dfc-56d1-eae7f3514d18@icir.org>
References: <alpine.DEB.2.20.1802091740370.3478@uplift.swm.pp.se>
	<7d7f8318-86ba-3dfc-56d1-eae7f3514d18@icir.org>
Message-ID: <alpine.DEB.2.20.1802092020540.3478@uplift.swm.pp.se>

On Fri, 9 Feb 2018, Christian Kreibich wrote:

> Unfortunately that's not what the mode flag does -- this flag is 
> primarily for labeling a session as it enters our database, to allow us 
> to identify sets of sessions that are coming in from a certain 
> experiment, setting, partner site, etc. It may also mean a slightly 
> tweaked set of tests, but there's currently no way to run a single test. 
> If you send us some of your session IDs (feel free to keep these 
> off-list), we'd be happy to take a look regarding what might be going 
> wrong.

Thanks for the response. I already have packet dumps of the test, so I can 
tell what the last packet is that is sent right before the device crashes 
(and I also know that it has to do with the packet forwarding acceleration 
code on the device, because when I unload all acceleration then things 
work as expected without any crash).

I know the device crash occurs around test 58/110, but I am not 100% sure 
exactly which of these tests that cause it, because it could also be that 
the crash (and it's a kernel crash+reboot) is caused by a delayed reaction 
to an earlier test.

So it would be really helpful if the CLI client would support to just run 
one of the tests, and perhaps not even create a report. Another option 
would be to be able to insert a 5 second delay between each test, to 
easily isolate them.

So if you're going to re-open the cli client for development, please take 
what I wrote above under advisement.

Thanks!

-- 
Mikael Abrahamsson    email: swmike at swm.pp.se

From christian at icir.org  Mon Feb 12 11:04:38 2018
From: christian at icir.org (Christian Kreibich)
Date: Mon, 12 Feb 2018 11:04:38 -0800
Subject: [Netalyzr] how to run single test in the CLI version
In-Reply-To: <alpine.DEB.2.20.1802092020540.3478@uplift.swm.pp.se>
References: <alpine.DEB.2.20.1802091740370.3478@uplift.swm.pp.se>
	<7d7f8318-86ba-3dfc-56d1-eae7f3514d18@icir.org>
	<alpine.DEB.2.20.1802092020540.3478@uplift.swm.pp.se>
Message-ID: <bbf011ba-b4af-b830-2c5e-21a894a8bd59@icir.org>

Hey Mikael,

On 02/09/2018 11:27 AM, Mikael Abrahamsson wrote:
> Thanks for the response. I already have packet dumps of the test, so I can
> tell what the last packet is that is sent right before the device crashes
> (and I also know that it has to do with the packet forwarding acceleration
> code on the device, because when I unload all acceleration then things
> work as expected without any crash).

Great! Let's take this off-list and troubleshoot further. I'll ping you 
shortly.

> I know the device crash occurs around test 58/110, but I am not 100% sure
> exactly which of these tests that cause it, because it could also be that
> the crash (and it's a kernel crash+reboot) is caused by a delayed reaction
> to an earlier test.

Yes, that's true. It's intriguing, because tests that "stress" the link 
heavily should only occur near the end of the session.

> So it would be really helpful if the CLI client would support to just run
> one of the tests, and perhaps not even create a report. Another option
> would be to be able to insert a 5 second delay between each test, to
> easily isolate them.
> 
> So if you're going to re-open the cli client for development, please take
> what I wrote above under advisement.

Duly noted!

Best,
Christian

From christian at icir.org  Mon Feb 12 11:08:52 2018
From: christian at icir.org (Christian Kreibich)
Date: Mon, 12 Feb 2018 11:08:52 -0800
Subject: [Netalyzr] Upcoming maintenance
Message-ID: <f411c9bc-fde4-4431-13b0-568a95a5f85f@icir.org>

Hi folks,

Tomorrow (Tuesday, Feb 13th) morning Pacific time we'll be conducting 
maintenance on the Netalyzr front-end server. During the service window, 
the service will be offline for clients. The outage will start around 
10am PT and shouldn't take more than 2h.

Thanks,
Christian

From narseo at ICSI.Berkeley.EDU  Thu Feb 15 01:39:11 2018
From: narseo at ICSI.Berkeley.EDU (Narseo Vallina Rodriguez)
Date: Thu, 15 Feb 2018 10:39:11 +0100
Subject: [Netalyzr] New Android app release
Message-ID: <CAJmR=Q=3sR7DVq=AANRustR9vais4zfp5GPdeTg=57DgbHfwNA@mail.gmail.com>

Dear Netalyzr users,

We are happy to announce that we finally fixed the issues present on
the Android app.

The reason for this misbehaviour was caused by a small number of tests
designed to measure CDN performance and CDN-ISP integration. In those
tests, we were relying on a number of small and popular objects hosted
in large CDNs which were suddenly removed around late 2017. As a
result, the tests never completed as many of you reported.

We removed this test to make sure that we remain in full control of
the tests. In addition to that, we also improved some of the tests to
provide a better experience. The new version (1.6) has been already
indexed by Google Play.

Thank you very much for your patience and for letting us know about
the problem. As always, your feedback is very useful to us.

In case that you report any other problem in the new version, we would
appreciate your input.

Do not hesitate to contact us if you have any question.

Sincerely,

-- 
Kind regards,

Narseo Vallina-Rodriguez
Research Staff, ICSI
Personal web: http://www.icsi.berkeley.edu/~narseo
Twitter: https://twitter.com/narseo