source: branches/samba-3.2.x/docs-xml/Samba3-ByExample/SBE-HighAvailability.xml

Last change on this file was 203, checked in by Herwig Bauernfeind, 16 years ago

Missing 3.2.2 client and HOWTO files
File size: 30.2 KB
RevLine 
[203]1<?xml version="1.0" encoding="iso-8859-1"?>
2<!DOCTYPE chapter PUBLIC "-//Samba-Team//DTD DocBook V4.2-Based Variant V1.0//EN" "http://www.samba.org/samba/DTD/samba-doc">
3
4<chapter id="HA">
5<title>Performance, Reliability, and Availability</title>
6
7 <para>
8 <indexterm><primary>performance</primary></indexterm>
9 <indexterm><primary>reliability</primary></indexterm>
10 <indexterm><primary>availability</primary></indexterm>
11 Well, you have reached one of the last chapters of this book. It is customary to attempt
12 to wrap up the theme and contents of a book in what is generally regarded as the
13 chapter that should draw conclusions. This book is a suspense thriller, and since
14 the plot of the stories told mostly lead you to bigger, better Samba-3 networking
15 solutions, it is perhaps appropriate to close this book with a few pertinent comments
16 regarding some of the things everyone can do to deliver a reliable Samba-3 network.
17 </para>
18
19 <blockquote><attribution>Anonymous</attribution><para>
20 In a world so full of noise, how can the sparrow be heard?
21 </para></blockquote>
22
23<sect1>
24 <title>Introduction</title>
25
26 <para>
27 <indexterm><primary>clustering</primary></indexterm>
28 The sparrow is a small bird whose sounds are drowned out by the noise of the busy
29 world it lives in. Likewise, the simple steps that can be taken to improve the
30 reliability and availability of a Samba network are often drowned out by the volume
31 of discussions about grandiose Samba clustering designs. This is not intended to
32 suggest that clustering is not important, because clearly it is. This chapter does not devote
33 itself to discussion of clustering because each clustering methodology uses its own
34 custom tools and methods. Only passing comments are offered concerning these methods.
35 </para>
36
37 <para>
38 <indexterm><primary>cluster</primary></indexterm>
39 <indexterm><primary>samba cluster</primary></indexterm>
40 <indexterm><primary>scalability</primary></indexterm>
41<ulink url="http://www.google.com/search?hl=en&amp;lr=&amp;ie=ISO-8859-1&amp;q=samba+cluster&amp;btnG=Google+Search">A search</ulink>
42 for <quote>samba cluster</quote> produced 71,600 hits. And a search for <quote>highly available samba</quote>
43 and <quote>highly available windows</quote> produced an amazing number of references.
44 It is clear from the resources on the Internet that Windows file and print services
45 availability, reliability, and scalability are of vital interest to corporate network users.
46 </para>
47
48 <para>
49 <indexterm><primary>performance</primary></indexterm>
50 So without further background, you can review a checklist of simple steps that
51 can be taken to ensure acceptable network performance while keeping costs of ownership
52 well under control.
53 </para>
54
55</sect1>
56
57<sect1>
58 <title>Dissection and Discussion</title>
59
60 <para>
61 <indexterm><primary>simple</primary></indexterm>
62 <indexterm><primary>complexities</primary></indexterm>
63 If it is your purpose to get the best mileage out of your Samba servers, there is one rule that
64 must be obeyed. If you want the best, keep your implementation as simple as possible. You may
65 well be forced to introduce some complexities, but you should do so only as a last resort.
66 </para>
67
68 <para>
69 Simple solutions are likely to be easier to get right than are complex ones. They certainly
70 make life easier for your successor. Simple implementations can be more readily audited than can
71 complex ones.
72 </para>
73
74 <para>
75 <indexterm><primary>broken behavior</primary></indexterm>
76 <indexterm><primary>poor performance</primary></indexterm>
77 Problems reported by users fall into three categories: configurations that do not work, those
78 that have broken behavior, and poor performance. The term <emphasis>broken behavior</emphasis>
79 means that the function of a particular Samba component appears to work sometimes, but not at
80 others. The resulting intermittent operation is clearly unacceptable. An example of
81 <emphasis>broken behavior</emphasis> known to many Windows networking users occurs when the
82 list of Windows machines in MS Explorer changes, sometimes listing machines that are running
83 and at other times not listing them even though the machines are in use on the network.
84 </para>
85
86 <para>
87 <indexterm><primary>smbfs</primary></indexterm>
88 <indexterm><primary>smbmnt</primary></indexterm>
89 <indexterm><primary>smbmount</primary></indexterm>
90 <indexterm><primary>smbumnt</primary></indexterm>
91 <indexterm><primary>smbumount</primary></indexterm>
92 <indexterm><primary>front-end</primary></indexterm>
93 A significant number of reports concern problems with the <command>smbfs</command> file system
94 driver that is part of the Linux kernel, not part of Samba. Users continue to interpret that
95 <command>smbfs</command> is part of Samba, simply because Samba includes the front-end tools
96 that are used to manage <command>smbfs</command>-based file service connections. So, just
97 for the record, the tools <command>smbmnt</command>, <command>smbmount</command>,
98 <command>smbumount</command>, and <command>smbumnt</command> are front-end
99 facilities to core drivers that are supplied as part of the Linux kernel. These tools share a
100 common infrastructure with some Samba components, but they are not maintained as part of
101 Samba and are really foreign to it.
102 </para>
103
104 <para>
105 <indexterm><primary>cifsfs</primary></indexterm>
106 The new project, <command>cifsfs</command>, is destined to replace <command>smbfs</command>.
107 It, too, is not part of Samba, even though one of the Samba Team members is a prime mover in
108 this project.
109 </para>
110
111 <para>
112 Table 13.1 lists typical causes of:
113 </para>
114
115 <itemizedlist>
116 <listitem><para>Not Working (NW)</para></listitem>
117 <listitem><para>Broken Behavior (BB)</para></listitem>
118 <listitem><para>Poor Performance (PP)</para></listitem>
119 </itemizedlist>
120
121
122 <table id="ProbList">
123 <title>Effect of Common Problems</title>
124 <tgroup cols="4">
125 <colspec align="left"/>
126 <colspec align="center"/>
127 <colspec align="center"/>
128 <colspec align="center"/>
129 <thead>
130 <row>
131 <entry><para>Problem</para></entry>
132 <entry><para>NW</para></entry>
133 <entry><para>BB</para></entry>
134 <entry><para>PP</para></entry>
135 </row>
136 </thead>
137 <tbody>
138 <row>
139 <entry><para>File locking</para></entry>
140 <entry><para>-</para></entry>
141 <entry><para>X</para></entry>
142 <entry><para>-</para></entry>
143 </row>
144 <row>
145 <entry><para>Hardware problems</para></entry>
146 <entry><para>X</para></entry>
147 <entry><para>X</para></entry>
148 <entry><para>X</para></entry>
149 </row>
150 <row>
151 <entry><para>Incorrect authentication</para></entry>
152 <entry><para>X</para></entry>
153 <entry><para>X</para></entry>
154 <entry><para>-</para></entry>
155 </row>
156 <row>
157 <entry><para>Incorrect configuration</para></entry>
158 <entry><para>X</para></entry>
159 <entry><para>X</para></entry>
160 <entry><para>X</para></entry>
161 </row>
162 <row>
163 <entry><para>LDAP problems</para></entry>
164 <entry><para>X</para></entry>
165 <entry><para>X</para></entry>
166 <entry><para>-</para></entry>
167 </row>
168 <row>
169 <entry><para>Name resolution</para></entry>
170 <entry><para>X</para></entry>
171 <entry><para>X</para></entry>
172 <entry><para>X</para></entry>
173 </row>
174 <row>
175 <entry><para>Printing problems</para></entry>
176 <entry><para>X</para></entry>
177 <entry><para>X</para></entry>
178 <entry><para>-</para></entry>
179 </row>
180 <row>
181 <entry><para>Slow file transfer</para></entry>
182 <entry><para>-</para></entry>
183 <entry><para>-</para></entry>
184 <entry><para>X</para></entry>
185 </row>
186 <row>
187 <entry><para>Winbind problems</para></entry>
188 <entry><para>X</para></entry>
189 <entry><para>X</para></entry>
190 <entry><para>-</para></entry>
191 </row>
192 </tbody>
193 </tgroup>
194 </table>
195
196 <para>
197 <indexterm><primary>network hygiene</primary></indexterm>
198 It is obvious to all that the first requirement (as a matter of network hygiene) is to eliminate
199 problems that affect basic network operation. This book has provided sufficient working examples
200 to help you to avoid all these problems.
201 </para>
202
203</sect1>
204
205<sect1>
206 <title>Guidelines for Reliable Samba Operation</title>
207
208 <para>
209 <indexterm><primary>resilient</primary></indexterm>
210 <indexterm><primary>extreme demand</primary></indexterm>
211 Your objective is to provide a network that works correctly, can grow at all times, is resilient
212 at times of extreme demand, and can scale to meet future needs. The following subject areas provide
213 pointers that can help you today.
214 </para>
215
216 <sect2>
217 <title>Name Resolution</title>
218
219 <para>
220 There are three basic current problem areas: bad hostnames, routed networks, and network collisions.
221 These are covered in the following discussion.
222 </para>
223
224 <sect3>
225 <title>Bad Hostnames</title>
226
227 <para>
228 <indexterm><primary>DHCP</primary><secondary>client</secondary></indexterm>
229 <indexterm><primary>netbios name</primary></indexterm>
230 <indexterm><primary>localhost</primary></indexterm>
231 <indexterm><primary>/etc/hosts</primary></indexterm>
232 <indexterm><primary>NetBIOS</primary></indexterm>
233 When configured as a DHCP client, a number of Linux distributions set the system hostname
234 to <constant>localhost</constant>. If the parameter <parameter>netbios name</parameter> is not
235 specified to something other than <constant>localhost</constant>, the Samba server appears
236 in the Windows Explorer as <constant>LOCALHOST</constant>. Moreover, the entry in the <filename>/etc/hosts</filename>
237 on the Linux server points to IP address <constant>127.0.0.1</constant>. This means that
238 when the Windows client obtains the IP address of the Samba server called <constant>LOCALHOST</constant>,
239 it obtains the IP address <constant>127.0.0.1</constant> and then proceeds to attempt to
240 set up a NetBIOS over TCP/IP connection to it. This cannot work, because that IP address is
241 the local Windows machine itself. Hostnames must be valid for Windows networking to function
242 correctly.
243 </para>
244
245 <para>
246 <indexterm><primary>digits</primary></indexterm>
247 A few sites have tried to name Windows clients and Samba servers with a name that begins
248 with the digits 1-9. This does not work either because it may result in the client or
249 server attempting to use that name as an IP address.
250 </para>
251
252 <para>
253 <indexterm><primary>DNS</primary><secondary>name lookup</secondary></indexterm>
254 <indexterm><primary>resolve</primary></indexterm>
255 A Samba server called <constant>FRED</constant> in a NetBIOS domain called <constant>COLLISION</constant>
256 in a network environment that is part of the fully-qualified Internet domain namespace known
257 as <constant>parrots.com</constant>, results in DNS name lookups for <constant>fred.parrots.com</constant>
258 and <constant>collision.parrots.com</constant>. It is therefore a mistake to name the domain
259 (workgroup) <constant>collision.parrots.com</constant>, since this results in DNS lookup
260 attempts to resolve <constant>fred.parrots.com.parrots.com</constant>, which most likely
261 fails given that you probably do not have this in your DNS namespace.
262 </para>
263
264 <note><para>
265 <indexterm><primary>Active Directory</primary><secondary>realm</secondary></indexterm>
266 <indexterm><primary>ADS</primary></indexterm>
267 <indexterm><primary>DNS</primary></indexterm>
268 An Active Directory realm called <constant>collision.parrots.com</constant> is perfectly okay,
269 although it too must be capable of being resolved via DNS, something that functions correctly
270 if Windows 200x ADS has been properly installed and configured.
271 </para></note>
272
273 </sect3>
274
275 <sect3>
276 <title>Routed Networks</title>
277
278 <para>
279 <indexterm><primary>NetBIOS</primary></indexterm>
280 <indexterm><primary>UDP</primary><secondary>broadcast</secondary></indexterm>
281 <indexterm><primary>broadcast</primary></indexterm>
282 NetBIOS networks (Windows networking with NetBIOS over TCP/IP enabled) makes extensive use
283 of UDP-based broadcast traffic, as you saw during the exercises in <link linkend="primer"/>.
284 </para>
285
286 <para>
287 <indexterm><primary>routers</primary></indexterm>
288 <indexterm><primary>forwarded</primary></indexterm>
289 <indexterm><primary>multi-subnet</primary></indexterm>
290 UDP broadcast traffic is not forwarded by routers. This means that NetBIOS broadcast-based
291 networking cannot function across routed networks (i.e., multi-subnet networks) unless
292 special provisions are made:
293 </para>
294
295 <itemizedlist>
296 <listitem><para>
297 <indexterm><primary>LMHOSTS</primary></indexterm>
298 <indexterm><primary>remote announce</primary></indexterm>
299 <indexterm><primary>remote browse sync</primary></indexterm>
300 Either install on every Windows client an LMHOSTS file (located in the directory
301 <filename>C:\windows\system32\drivers\etc</filename>). It is also necessary to
302 add to the Samba server &smb.conf; file the parameters <parameter>remote announce</parameter>
303 and <parameter>remote browse sync</parameter>. For more information, refer to the online
304 manual page for the &smb.conf; file.
305 </para></listitem>
306
307 <listitem><para>
308 <indexterm><primary>WINS</primary><secondary>server</secondary></indexterm>
309 Or configure Samba as a WINS server, and configure all network clients to use that
310 WINS server in their TCP/IP configuration.
311 </para></listitem>
312 </itemizedlist>
313
314 <note><para>
315 <indexterm><primary>WINS</primary><secondary>name resolution</secondary></indexterm>
316 <indexterm><primary>DNS</primary></indexterm>
317 The use of DNS is not an acceptable substitute for WINS. DNS does not store specific
318 information regarding NetBIOS networking particulars that get stored in the WINS
319 name resolution database and that Windows clients require and depend on.
320 </para></note>
321
322 </sect3>
323
324 <sect3>
325 <title>Network Collisions</title>
326
327 <para>
328 <indexterm><primary>network</primary><secondary>collisions</secondary></indexterm>
329 <indexterm><primary>network</primary><secondary>timeouts</secondary></indexterm>
330 <indexterm><primary>collision rates</primary></indexterm>
331 <indexterm><primary>network</primary><secondary>load</secondary></indexterm>
332 Excessive network activity causes NetBIOS network timeouts. Timeouts may result in
333 blue screen of death (BSOD) experiences. High collision rates may be caused by excessive
334 UDP broadcast activity, by defective networking hardware, or through excessive network
335 loads (another way of saying that the network is poorly designed).
336 </para>
337
338 <para>
339 The use of WINS is highly recommended to reduce network broadcast traffic, as outlined
340 in <link linkend="primer"/>.
341 </para>
342
343 <para>
344 <indexterm><primary>netbios forwarding</primary></indexterm>
345 <indexterm><primary>broadcast storms</primary></indexterm>
346 <indexterm><primary>performance</primary></indexterm>
347 Under no circumstances should the facility be supported by many routers, known as <constant>NetBIOS
348 forwarding</constant>, unless you know exactly what you are doing. Inappropriate use of this
349 facility can result in UDP broadcast storms. In one case in 1999, a university network became
350 unusable due to NetBIOS forwarding being enabled on all routers. The problem was discovered during performance
351 testing of a Samba server. The maximum throughput on a 100-Base-T (100 MB/sec) network was
352 less than 15 KB/sec. After the NetBIOS forwarding was turned off, file transfer performance
353 immediately returned to 11 MB/sec.
354 </para>
355
356 </sect3>
357
358 </sect2>
359
360 <sect2>
361 <title>Samba Configuration</title>
362
363 <para>
364 As a general rule, the contents of the &smb.conf; file should be kept as simple as possible.
365 No parameter should be specified unless you know it is essential to operation.
366 </para>
367
368 <para>
369 <indexterm><primary>document the settings</primary></indexterm>
370 <indexterm><primary>documented</primary></indexterm>
371 <indexterm><primary>optimized</primary></indexterm>
372 Many UNIX administrators like to fully document the settings in the &smb.conf; file. This is a
373 bad idea because it adds content to the file. The &smb.conf; file is re-read by every <command>smbd</command>
374 process every time the file timestamp changes (or, on systems where this does not work, every 20 seconds or so).
375 </para>
376
377 <para>
378 As the size of the &smb.conf; file grows, the risk of introducing parsing errors also increases.
379 It is recommended to keep a fully documented &smb.conf; file on hand, and then to operate Samba only
380 with an optimized file.
381 </para>
382
383 <para><indexterm>
384 <primary>testparm</primary>
385 </indexterm>
386 The preferred way to maintain a documented file is to call it something like <filename>smb.conf.master</filename>.
387 You can generate the optimized file by executing:
388<screen>
389&rootprompt; testparm -s smb.conf.master > smb.conf
390</screen>
391 You should carefully observe all warnings issued. It is also a good practice to execute the following
392 command to confirm correct interpretation of the &smb.conf; file contents:
393<screen>
394&rootprompt; testparm
395Load smb config files from /etc/samba/smb.conf
396Can't find include file /etc/samba/machine.
397Processing section "[homes]"
398Processing section "[print$]"
399Processing section "[netlogon]"
400Processing section "[Profiles]"
401Processing section "[printers]"
402Processing section "[media]"
403Processing section "[data]"
404Processing section "[cdr]"
405Processing section "[apps]"
406Loaded services file OK.
407'winbind separator = +' might cause problems with group membership.
408Server role: ROLE_DOMAIN_PDC
409Press enter to see a dump of your service definitions
410</screen>
411 <indexterm><primary>fatal problem</primary></indexterm>
412 You now, of course, press the enter key to complete the command, or else abort it by pressing Ctrl-C.
413 The important thing to note is the noted Server role, as well as warning messages. Noted configuration
414 conflicts must be remedied before proceeding. For example, the following error message represents a
415 common fatal problem:
416<screen>
417ERROR: both 'wins support = true' and 'wins server = &lt;server list&gt;'
418cannot be set in the smb.conf file. nmbd will abort with this setting.
419</screen>
420 </para>
421
422 <para>
423 <indexterm><primary>performance degradation</primary></indexterm>
424 <indexterm><primary>socket options</primary></indexterm>
425 <indexterm><primary>socket address</primary></indexterm>
426 There are two parameters that can cause severe network performance degradation: <parameter>socket options</parameter>
427 and <parameter>socket address</parameter>. The <parameter>socket options</parameter> parameter was often necessary
428 when Samba was used with the Linux 2.2.x kernels. Later kernels are largely self-tuning and seldom benefit from
429 this parameter being set. Do not use either parameter unless it has been proven necessary to use them.
430 </para>
431
432 <para>
433 <indexterm><primary>strict sync</primary></indexterm>
434 <indexterm><primary>sync always</primary></indexterm>
435 <indexterm><primary>severely degrade</primary></indexterm>
436 <indexterm><primary>network</primary><secondary>performance</secondary></indexterm>
437 Another &smb.conf; parameter that may cause severe network performance degradation is the
438 <parameter>strict sync</parameter> parameter. Do not use this at all. There is no good reason
439 to use this with any modern Windows client. The <parameter>strict sync</parameter> is often
440 used with the <parameter>sync always</parameter> parameter. This, too, can severely
441 degrade network performance, so do not set it; if you must, do so with caution.
442 </para>
443
444 <para>
445 <indexterm><primary>opportunistic locking</primary></indexterm>
446 <indexterm><primary>file caching</primary></indexterm>
447 <indexterm><primary>caching</primary></indexterm>
448 <indexterm><primary>oplocks</primary></indexterm>
449 Finally, many network administrators deliberately disable opportunistic locking support. While this
450 does not degrade Samba performance, it significantly degrades Windows client performance because
451 this disables local file caching on Windows clients and forces every file read and written to
452 invoke a network read or write call. If for any reason you must disable oplocks (opportunistic locking)
453 support, do so only on the share on which it is required. That way, all other shares can provide
454 oplock support for operations that are tolerant of it. See <link linkend="ch12dblck"/> for more
455 information.
456 </para>
457
458 </sect2>
459
460 <sect2>
461 <title>Use and Location of BDCs</title>
462
463 <para>
464 <indexterm><primary>BDC</primary></indexterm>
465 <indexterm><primary>PDC</primary></indexterm>
466 <indexterm><primary>routed network</primary></indexterm>
467 <indexterm><primary>wide-area network</primary></indexterm>
468 <indexterm><primary>network segment</primary></indexterm>
469 On a network segment where there is a PDC and a BDC, the BDC carries the bulk of the network logon
470 processing. If the BDC is a heavily loaded server, the PDC carries a greater proportion of
471 authentication and logon processing. When a sole BDC on a routed network segment gets heavily
472 loaded, it is possible that network logon requests and authentication requests may be directed
473 to a BDC on a distant network segment. This significantly hinders WAN operations
474 and is undesirable.
475 </para>
476
477 <para>
478 <indexterm><primary>Domain Member</primary></indexterm>
479 <indexterm><primary>Domain Controller</primary></indexterm>
480 As a general guide, instead of adding domain member servers to a network, you would be better advised
481 to add BDCs until there are fewer than 30 Windows clients per BDC. Beyond that ratio, you should add
482 domain member servers. This practice ensures that there are always sufficient domain controllers
483 to handle logon requests and authentication traffic.
484 </para>
485
486 </sect2>
487
488 <sect2>
489 <title>Use One Consistent Version of MS Windows Client</title>
490
491 <para>
492 Every network client has its own peculiarities. From a management perspective, it is easier to deal
493 with one version of MS Windows that is maintained to a consistent update level than it is to deal
494 with a mixture of clients.
495 </para>
496
497 <para>
498 On a number of occasions, particular Microsoft service pack updates of a Windows server or client
499 have necessitated special handling from the Samba server end. If you want to remain sane, keep you
500 client workstation configurations consistent.
501 </para>
502
503 </sect2>
504
505 <sect2>
506 <title>For Scalability, Use SAN-Based Storage on Samba Servers</title>
507
508 <para>
509 <indexterm><primary>SAN</primary></indexterm>
510 <indexterm><primary>synchronization</primary></indexterm>
511 Many SAN-based storage systems permit more than one server to share a common data store.
512 Use of a shared SAN data store means that you do not need to use time- and resource-hungry data
513 synchronization techniques.
514 </para>
515
516 <para>
517 <indexterm><primary>load distribution</primary></indexterm>
518 <indexterm><primary>clustering</primary></indexterm>
519 The use of a collection of relatively low-cost front-end Samba servers that are coupled to
520 a shared backend SAN data store permits load distribution while containing costs below that
521 of installing and managing a complex clustering facility.
522 </para>
523
524 </sect2>
525
526 <sect2>
527 <title>Distribute Network Load with MSDFS</title>
528
529 <para>
530 <indexterm><primary>MSDFS</primary></indexterm>
531 <indexterm><primary>distributed</primary></indexterm>
532 Microsoft DFS (distributed file system) technology has been implemented in Samba. MSDFS permits
533 data to be accessed from a single share and yet to actually be distributed across multiple actual
534 servers. Refer to <emphasis>TOSHARG2</emphasis>, Chapter 19, for information regarding
535 implementation of an MSDFS installation.
536 </para>
537
538 <para>
539 <indexterm><primary>front-end</primary><secondary>server</secondary></indexterm>
540 <indexterm><primary>MSDFS</primary></indexterm>
541 The combination of multiple backend servers together with a front-end server and use of MSDFS
542 can achieve almost the same as you would obtain with a clustered Samba server.
543 </para>
544
545 </sect2>
546
547 <sect2>
548 <title>Replicate Data to Conserve Peak-Demand Wide-Area Bandwidth</title>
549
550 <para>
551 <indexterm><primary>replicate</primary></indexterm>
552 <indexterm><primary>rsync</primary></indexterm>
553 <indexterm><primary>wide-area network</primary></indexterm>
554 Consider using <command>rsync</command> to replicate data across the WAN during times
555 of low utilization. Users can then access the replicated data store rather than needing to do so
556 across the WAN. This works best for read-only data, but with careful planning can be
557 implemented so that modified files get replicated back to the point of origin. Be careful with your
558 implementation if you choose to permit modification and return replication of the modified file;
559 otherwise, you may inadvertently overwrite important data.
560 </para>
561
562 </sect2>
563
564 <sect2>
565 <title>Hardware Problems</title>
566
567 <para>
568 <indexterm><primary>hardware prices</primary></indexterm>
569 <indexterm><primary>hardware problems</primary></indexterm>
570 <indexterm><primary>NICs</primary></indexterm>
571 <indexterm><primary>defective</primary><secondary>HUBs</secondary></indexterm>
572 <indexterm><primary>defective</primary><secondary>switches</secondary></indexterm>
573 <indexterm><primary>defective</primary><secondary>cables</secondary></indexterm>
574 Networking hardware prices have fallen sharply over the past 5 years. A surprising number
575 of Samba networking problems over this time have been traced to defective network interface
576 cards (NICs) or defective HUBs, switches, and cables.
577 </para>
578
579 <para>
580 <indexterm><primary>corrective action</primary></indexterm>
581 Not surprising is the fact that network administrators do not like to be shown to have made
582 a bad decision. Money saved in buying low-cost hardware may result in high costs incurred
583 in corrective action.
584 </para>
585
586 <para>
587 <indexterm><primary>intermittent</primary></indexterm>
588 <indexterm><primary>data corruption</primary></indexterm>
589 <indexterm><primary>slow network</primary></indexterm>
590 <indexterm><primary>low performance</primary></indexterm>
591 <indexterm><primary>data integrity</primary></indexterm>
592 Defective NICs, HUBs, and switches may appear as intermittent network access problems, intermittent
593 or persistent data corruption, slow network throughput, low performance, or even as BSOD
594 problems with MS Windows clients. In one case, a company updated several workstations with newer, faster
595 Windows client machines that triggered problems during logon as well as data integrity problems on
596 an older PC that was unaffected so long as the new machines were kept shut down.
597 </para>
598
599 <para>
600 Defective hardware problems may take patience and persistence before the real cause can be discovered.
601 </para>
602
603 <para>
604 <indexterm><primary>RAID controllers</primary></indexterm>
605 Networking hardware defects can significantly impact perceived Samba performance, but defective
606 RAID controllers as well as SCSI and IDE hard disk controllers have also been known to impair Samba server
607 operations. One business came to this realization only after replacing a Samba installation with MS
608 Windows Server 2000 running on the same hardware. The root of the problem completely eluded the network
609 administrator until the entire server was replaced. While you may well think that this would never
610 happen to you, experience shows that given the right (unfortunate) circumstances, this can happen to anyone.
611 </para>
612
613 </sect2>
614
615 <sect2>
616 <title>Large Directories</title>
617
618 <para>
619 There exist applications that create or manage directories containing many thousands of files. Such
620 applications typically generate many small files (less than 100 KB). At the best of times, under UNIX,
621 listing of the files in a directory that contains many files is slow. By default, Windows NT, 200x,
622 and XP Pro cause network file system directory lookups on a Samba server to be performed for both
623 the case preserving file name as well as for the mangled (8.3) file name. This incurs a huge overhead
624 on the Samba server that may slow down the system dramatically.
625 </para>
626
627 <para>
628 In an extreme case, the performance impact was dramatic. File transfer from the Samba server to a Windows
629 XP Professional workstation over 1 Gigabit Ethernet for 250-500 KB files was measured at approximately
630 30 MB/sec. But when tranferring a directory containing 120,000 files, all from 50KB to 60KB in size, the
631 transfer rate to the same workstation was measured at approximately 1.5 KB/sec. The net transfer was
632 on the order of a factor of 20-fold slower.
633 </para>
634
635 <para>
636 The symptoms that will be observed on the Samba server when a large directory is accessed will be that
637 aggregate I/O (typically blocks read) will be relatively low, yet the wait I/O times will be incredibly
638 long while at the same time the read queue is large. Close observation will show that the hard drive
639 that the file system is on will be thrashing wildly.
640 </para>
641
642 <para>
643 Samba-3.0.12 and later, includes new code that radically improves Samba perfomance. The secret to this is
644 really in the <smbconfoption name="case sensitive">True</smbconfoption> line. This tells smbd never to scan
645 for case-insensitive versions of names. So if an application asks for a file called <filename>FOO</filename>,
646 and it can not be found by a simple stat call, then smbd will return "file not found" immediately without
647 scanning the containing directory for a version of a different case.
648 </para>
649
650 <para>
651 Canonicalize all the files in the directory to have one case, upper or lower - either will do. Then set up
652 a new custom share for the application as follows:
653 <screen>
654 [bigshare]
655 path = /data/xrayfiles/neurosurgeons/
656 read only = no
657 case sensitive = True
658 default case = upper
659 preserve case = no
660 short preserve case = no
661 </screen>
662 </para>
663
664 <para>
665 All files and directories under the <parameter>path</parameter> directory must be in the same case
666 as specified in the &smb.conf; stanza. This means that smbd will not be able to find lower case
667 filenames with these settings. Note, this is done on a per-share basis.
668 </para>
669
670 </sect2>
671
672</sect1>
673
674<sect1>
675 <title>Key Points Learned</title>
676
677 <para>
678 This chapter has touched in broad sweeps on a number of simple steps that can be taken
679 to ensure that your Samba network is resilient, scalable, and reliable, and that it
680 performs well.
681 </para>
682
683 <para>
684 Always keep in mind that someone is responsible to maintain and manage your design.
685 In the long term, that may not be you. Spare a thought for your successor and give him or
686 her an even break.
687 </para>
688
689 <para>
690 <indexterm><primary>assumptions</primary></indexterm>
691 Last, but not least, you should not only keep the network design simple, but also be sure it is
692 well documented. This book may serve as your pattern for documenting every
693 aspect of your design, its implementation, and particularly the objects and assumptions
694 that underlie it.
695 </para>
696
697</sect1>
698
699
700</chapter>
701
Note: See TracBrowser for help on using the repository browser.