s3: Fix the read&x offset within a chain

[Samba/gebeck_regimport.git] / docs-xml / Samba3-HOWTO / TOSHARG-Speed.xml

<?xml version="1.0" encoding="iso-8859-1"?>
<!DOCTYPE chapter PUBLIC "-//Samba-Team//DTD DocBook V4.2-Based Variant V1.0//EN" "http://www.samba.org/samba/DTD/samba-doc">
<chapter id="speed">

<chapterinfo>
        <author>
                <firstname>Paul</firstname><surname>Cochrane</surname>
                <affiliation>
                        <orgname>Dundee Limb Fitting Centre</orgname>
                        <address><email>paulc@dth.scot.nhs.uk</email></address>
                </affiliation>
        </author>
        &author.jelmer;
        &author.jht;
</chapterinfo>

<title>Samba Performance Tuning</title>

<sect1>
<title>Comparisons</title>

<para>
The Samba server uses TCP to talk to the client, so if you are
trying to see if it performs well, you should really compare it to
programs that use the same protocol. The most readily available
programs for file transfer that use TCP are ftp or another TCP-based
SMB server.
</para>

<para>
If you want to test against something like an NT or Windows for Workgroups server, then
you will have to disable all but TCP on either the client or
server. Otherwise, you may well be using a totally different protocol
(such as NetBEUI) and comparisons may not be valid.
</para>

<para>
Generally, you should find that Samba performs similarly to ftp at raw
transfer speed. It should perform quite a bit faster than NFS,
although this depends on your system.
</para>

<para>
Several people have done comparisons between Samba and Novell, NFS, or
Windows NT. In some cases Samba performed the best, in others the worst. I
suspect the biggest factor is not Samba versus some other system, but the
hardware and drivers used on the various systems. Given similar
hardware, Samba should certainly be competitive in speed with other
systems.
</para>

</sect1>

<sect1>
<title>Socket Options</title>

<para>
There are a number of socket options that can greatly affect the
performance of a TCP-based server like Samba.
</para>

<para>
The socket options that Samba uses are settable both on the command
line with the <option>-O</option> option and in the &smb.conf; file.
</para>

<para>
The <smbconfoption name="socket options"/> section of the &smb.conf; manual page describes how
to set these and gives recommendations.
</para>

<para>
Getting the socket options correct can make a big difference to your
performance, but getting them wrong can degrade it by just as
much. The correct settings are very dependent on your local network.
</para>

<para>
The socket option TCP_NODELAY is the one that seems to make the biggest single difference
for most networks. Many people report that adding
<smbconfoption name="socket options">TCP_NODELAY</smbconfoption>
doubles the read performance of a Samba drive. The best explanation I have seen for
this is that the Microsoft TCP/IP stack is slow in sending TCP ACKs.
</para>

<para>
There have been reports that setting <parameter>socket options = SO_RCVBUF=8192</parameter> in smb.conf
can seriously degrade Samba performance on the loopback adaptor (IP Address 127.0.0.1). It is strongly
recommended that before specifying any settings for <parameter>socket options</parameter>, the effect
first be quantitatively measured on the server being configured.
</para>

</sect1>

<sect1>
<title>Read Size</title>

<para>
The option <smbconfoption name="read size"/> affects the overlap of disk
reads/writes with network reads/writes. If the amount of data being
transferred in several of the SMB commands (currently SMBwrite, SMBwriteX, and
SMBreadbraw) is larger than this value, then the server begins writing
the data before it has received the whole packet from the network, or
in the case of SMBreadbraw, it begins writing to the network before
all the data has been read from disk.
</para>

<para>
This overlapping works best when the speeds of disk and network access
are similar, having little effect when the speed of one is much
greater than the other.
</para>

<para>
The default value is 16384, but little experimentation has been
done as yet to determine the optimal value, and it is likely that the best
value will vary greatly between systems anyway. A value over 65536 is
pointless and will cause you to allocate memory unnecessarily.
</para>

</sect1>

<sect1>
<title>Max Xmit</title>

<para>
        At startup the client and server negotiate a <parameter>maximum transmit</parameter> size,
which limits the size of nearly all SMB commands. You can set the
maximum size that Samba will negotiate using the <smbconfoption name="max xmit"/> option
in &smb.conf;. Note that this is the maximum size of SMB requests that 
Samba will accept, but not the maximum size that the client will accept.
The client maximum receive size is sent to Samba by the client, and Samba
honors this limit.
</para>

<para>
It defaults to 65536 bytes (the maximum), but it is possible that some
clients may perform better with a smaller transmit unit. Trying values
of less than 2048 is likely to cause severe problems.
In most cases the default is the best option.
</para>

</sect1>

<sect1>
<title>Log Level</title>

<para>
If you set the log level (also known as <smbconfoption name="debug level"/>) higher than 2,
then you may suffer a large drop in performance. This is because the
server flushes the log file after each operation, which can be quite
expensive. 
</para>
</sect1>

<sect1>
<title>Read Raw</title>

<para>
The <smbconfoption name="read raw"/> operation is designed to be an optimized, low-latency
file read operation. A server may choose to not support it,
however, and Samba makes support for <smbconfoption name="read raw"/> optional, with it
being enabled by default.
</para>

<para>
In some cases clients do not handle <smbconfoption name="read raw"/> very well and actually
get lower performance using it than they get using the conventional
read operations, so you might like to try <smbconfoption name="read raw">no</smbconfoption> and see what happens on your
network. It might lower, raise, or not affect your performance. Only
testing can really tell.
</para>

</sect1>

<sect1>
<title>Write Raw</title>

<para>
The <smbconfoption name="write raw"/> operation is designed to be an optimized, low-latency
file write operation. A server may choose to not support it, however, and Samba makes support for
<smbconfoption name="write raw"/> optional, with it being enabled by default.
</para>

<para>
Some machines may find <smbconfoption name="write raw"/> slower than normal write, in which
case you may wish to change this option.
</para>

</sect1>

<sect1>
<title>Slow Logins</title>

<para>
Slow logins are almost always due to the password checking time. Using
the lowest practical <smbconfoption name="password level"/> will improve things. 
</para>

</sect1>

<sect1>
<title>Client Tuning</title>

<para>
Often a speed problem can be traced to the client. The client (for
example Windows for Workgroups) can often be tuned for better TCP
performance. Check the sections on the various clients in 
<link linkend="Other-Clients">Samba and Other CIFS Clients</link>.
</para>

</sect1>

<sect1>
<title>Samba Performance Problem Due to Changing Linux Kernel</title>

<para>
A user wrote the following to the mailing list:
</para>

<blockquote>
<para>
<indexterm><primary>Gentoo</primary></indexterm>
<indexterm><primary>slow network</primary></indexterm>
I am running Gentoo on my server and Samba 2.2.8a. Recently I changed kernel versions from
<filename>linux-2.4.19-gentoo-r10</filename> to <filename>linux-2.4.20-wolk4.0s</filename>. Now I have a
performance issue with Samba.  Many of you will probably say,  <quote>Move to vanilla sources!</quote> Well, I
tried that and it didn't work. I have a 100MB LAN and two computers (Linux and Windows 2000). The Linux server
shares directories with DivX files, the client (Windows 2000) plays them via LAN. Before, when I was running
the 2.4.19 kernel, everything was fine, but now movies freeze and stop. I tried moving files between the
server and Windows, and it is terribly slow.
</para>
</blockquote>

<para>
The answer he was given is:
</para>

<blockquote>
<para>
<indexterm><primary>ifconfig</primary></indexterm>
<indexterm><primary>framing error</primary></indexterm>
<indexterm><primary>collisions</primary></indexterm>
Grab the mii-tool and check the duplex settings on the NIC.  My guess is that it is a link layer issue, not an
application layer problem. Also run ifconfig and verify that the framing error, collisions, and so on, look
normal for ethernet.
</para>
</blockquote>

</sect1>

<sect1>
<title>Corrupt tdb Files</title>

<para>
<indexterm><primary>PDC</primary></indexterm>
<indexterm><primary>mbd kept spawning</primary></indexterm>
<indexterm><primary>/var/locks/*.tdb</primary></indexterm>
Our Samba PDC server has been hosting three TB of data to our 500+ users [Windows NT/XP]  for the last three
years using Samba without a problem.  Today all shares went very slow. Also, the main smbd kept spawning new
processes, so we had 1600+ running SMDB's (normally we average 250).  It crashed the SUN E3500 cluster twice.
After a lot of searching, I decided to <command>rm /var/locks/*.tdb</command>. Happy again.
</para>

<para>
<emphasis>Question:</emphasis> Is there any method of keeping the *.tdb files in top condition, or
how can I detect early corruption?
</para>

<para>
<indexterm><primary>tdbbackup</primary></indexterm>
<indexterm><primary>nmbd</primary></indexterm>
<emphasis>Answer:</emphasis> Yes, run <command>tdbbackup</command> each time after stopping nmbd and before starting nmbd.
</para>

<para>
<emphasis>Question:</emphasis> What I also would like to mention is that the service latency seems
a lot lower than before the locks cleanup. Any ideas on keeping it top notch?
</para>

<para>
<emphasis>Answer:</emphasis> Yes. Same answer as for previous question!
</para>

</sect1>

<sect1>
<title>Samba Performance is Very Slow</title>

<para>
<indexterm><primary>slow performance</primary></indexterm>
A site reported experiencing very baffling symptoms with MYOB Premier opening and
accessing its data files. Some  operations on the file would take between 40 and
45 seconds.
</para>

<para>
<indexterm><primary>printer monitor</primary></indexterm>
<indexterm><primary>pauses</primary></indexterm>
It turned out that the printer monitor program running on the Windows
clients was causing the problems. From the logs, we saw activity coming
through with pauses of about 1 second.
</para>

<para>
<indexterm><primary>networks access</primary></indexterm>
<indexterm><primary>printing now</primary></indexterm>
Stopping the monitor software resulted in the networks access at normal
(quick) speed. Restarting the program caused the speed to slow down
again. The printer was a Canon LBP-810 and the relevant task was
something like CAPON (not sure on spelling). The monitor software
displayed a "printing now" dialog on the client during printing.
</para>

<para>
We discovered this by starting with a clean install of Windows and
trying the application at every step of the installation of other software
process (we had to do this many times).
</para>

<para>
Moral of the story: Check everything (other software included)!
</para>

</sect1>

</chapter>