Prev: [Samba] Windows 7 + Samba 3.4.5 locking problem
Next: [Samba] unable to access samba server from long distance
From: Lars Bensmann on 5 Mar 2010 22:50
On Fri, Mar 05, 2010 at 05:16:10PM -0800, Jeremy Allison wrote:
> Did you try setting:
>
> kernel oplocks = no
> oplocks = no
> level2 oplocks = no
> posix locking = no

I was not sure if I tried all your settings at once, so I tested it just
now. (I put kernel oplocks = no as a global option as it had no effect on
a share only basis according to testparm.)

> on that share ? This will leave you with only Samba share mode
> and byte range locking (no mapping down to the underlying
> system) - which should behave identically to Windows.

Unfortunately it did not. Already the first client experienced the delay.

> Remeber to restart Samba before testing.

I did.

Can I gather any more relevant data to find out what goes wrong?

Lars

--
The eagle may soar, but the weasel never gets sucked into a jet engine.
--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/options/samba
From: Volker Lendecke on 6 Mar 2010 02:00
On Sat, Mar 06, 2010 at 04:36:44AM +0100, Lars Bensmann wrote:
> > Remeber to restart Samba before testing.
>
> I did.
>
> Can I gather any more relevant data to find out what goes wrong?

Does the debug level you set on Samba make any difference?
debug level = 0 is definitely faster than debug level = 10.

We need network traces of the app starting against Windows
and then the same against Samba. They are going to be huge
and it's going to be difficult to debug, but it is the only
way.

http://wiki.samba.org/index.php/Capture_Packets

Volker
--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/options/samba
From: Helmut Hullen on 6 Mar 2010 03:00
Hallo, Lars,

Du meintest am 05.03.10:

> The application works with a lot of database files (*.DBF) and
> everything works fine if only one workstation has started the
> application. Getting from the main menu to the data for the first
> patient takes about 3 to 5 seconds which is fine. But every station
> started after the first one takes about 35 seconds to get to the data
> for the first patient. Once the application is started further
> requests are fast.

Have you tested the switch too?
Switches, NICs, cable outlets ("Steckdosen") and cables may produce such
a behaviour too.

Viele Gruesse!
Helmut
--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/options/samba
From: Lars Bensmann on 6 Mar 2010 08:40
Hallo Helmut,

On Sat, Mar 06, 2010 at 08:22:00AM +0100, Helmut Hullen wrote:
> Have you tested the switch too?

not explicitely, but I highly doubt there is something wrong with it or
the cableing. I would be a major coincidence if they started making
trouble right at the moment when I upgraded the computers and the
application.

But to be on the safe side I just tested transfering a big file from the
server to two clients at the same time and the transfer speed was very
good.

Bye,
Lars

--
I browse at 14 pt and want to print at 12 pt. (That's just an example,
real men always use the smallest font available.)
-- Paul Phillips - www.mozilla.org
--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/options/samba
From: Lars Bensmann on 6 Mar 2010 09:00
On Sat, Mar 06, 2010 at 07:22:06AM +0100, Volker Lendecke wrote:
> On Sat, Mar 06, 2010 at 04:36:44AM +0100, Lars Bensmann wrote:
> > > Remeber to restart Samba before testing.
> >
> > I did.
> >
> > Can I gather any more relevant data to find out what goes wrong?
>
> Does the debug level you set on Samba make any difference?
> debug level = 0 is definitely faster than debug level = 10.

I just set debug level 10 for specific tests. Afterwards I set it back to
debug level 2. I guess that should not make such a big difference.

> We need network traces of the app starting against Windows
> and then the same against Samba. They are going to be huge
> and it's going to be difficult to debug, but it is the only
> way.

I already captured two traces against Samba. One for the first client
which is fast:
http://almosthappy.de/duhLd6Tm1GQoC3j0YyiJ637IS/capture.fast (16MB)
Then a capture of the second client which takes 30 seconds longer:
http://almosthappy.de/duhLd6Tm1GQoC3j0YyiJ637IS/capture.slow (26MB)

And here are two new traces. One for the first client against Windows 7
serving the files:
http://almosthappy.de/duhLd6Tm1GQoC3j0YyiJ637IS/capture.win7.first_client (11MB)
And then for the second client (although it does not have a noticable
performance impact):
http://almosthappy.de/duhLd6Tm1GQoC3j0YyiJ637IS/capture.win7.second_client (13MB)

Unfortunately, I have no clue what I should be looking for in these
traces.

Lars

--
When in danger or in doubt...
run in circles, scream, and shout !!!!!
--
To unsubscribe from this list go to the following URL and read the
instructions: https://lists.samba.org/mailman/options/samba
First  |  Prev  |  Next  |  Last
Pages: 1 2 3
Prev: [Samba] Windows 7 + Samba 3.4.5 locking problem
Next: [Samba] unable to access samba server from long distance