![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to scsi.ms CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [Research Unix] / researchv10dc / cmd / oworm / scsi|
Return to scsi.ms CVS log | Up to [Research Unix] / researchv10dc / cmd / oworm / scsi |
1.1 root 1: .so /usr/lib/tmac/tmac.s
2: .so /usr/lib/tmac/tmac.cs
3: .tr ~
4: .\" .RP here gives release paper (no cs)
5: .\" following arg should be removed so itds can get their jollies
6: .TI
7: Using the SONY WORM drive and jukebox
8: .AH "Andrew Hume" MH 11271 6262 2C-471 research!andrew ""
9: .SA
10: .FP palatino
11: This note describes the software available for handling
12: Write-Once-Read-Many (WORM) optical disks.
13: Some of the software is applicable to most devices with a SCSI interface.
14: .SE
15: .KW "optical disk" "SCSI" "" ""
16: .TY IM y
17: .NU 11271 880913 01 39199-11 ""
18: .MY "" "" y "" "" "" ""
19: .PR 0
20: .RL y
21: .CO y
22: Paul Glick
23: Ted Kowalski
24: Marty Shannon
25: M J Sheehan
26: E J Sitar
27: .CE
28: .CV y
29: .CE
30: .SC 11
31: .SH
32: .FP palatino
33: Introduction
34: .de CS
35: .DS
36: .nf
37: .ft CW
38: .ps 8
39: .vs 9.5
40: .ta 8n +8n +8n +8n +8n
41: .in 10n
42: ..
43: .de CE
44: .DE
45: ..
46: .PP
47: Optical drives are quite trendy nowadays (mid 1988).
48: The AT&T Bell Laboratories Computation Centers are doing
49: cost and performance evaluations of available hardware.
50: There even plans for an official AT&T file backup product based on WORMs.
51: For now and the next several years, WORM disks are the dominant
52: optical disk technology.
53: .PP
54: WORM disks range in size from 3.5 to 13 inches in diameter
55: and have capacities between a few hundred megabytes to a few gigabytes.
56: They can either be double-sided or single-sided
57: (but in general drives are single-sided;
58: double-sided disks must be ejected and then turned over
59: in order to access the second side).
60: There is also a tradeoff between capacity and speed of access.
61: Disks which spin at a constant angular velocity (CAV) are much faster
62: (a factor of 2\-3) at moving the heads to a given spot than disks
63: which spin at a constant linear velocity at the reading head (CLV).
64: On the other hand, the CLV disks typically have a 50% greater capacity
65: than the CAV disks.
66: All other characteristics are about the same, including transfer rate.
67: For example, the disks we use are
68: .TS
69: center;
70: c l.
71: disk type WDM-3DL0 (CLV)
72: disk size 12in
73: disk capacity 3.2GB (1.6GB/side)
74: seek time 600ms (1/3 stroke?)
75: read rate ~200KB/s
76: write rate ~75KB/s
77: .TE
78: I/O rates are measured on a multi-user timesharing system;
79: dedicated systems can do about 40% better.
80: .PP
81: Most WORM disks interface to computers by the SCSI bus.
82: Our computing environment (VAXen) doesn't support SCSI directly;
83: we use SCSI controllers that attach to the Unibus.
84: This is not as bad as it seems because for once, DEC had a good idea and
85: designed and implemented a device independent protocol
86: (Mass Storage Control Protocol or MSCP) for its mass storage devices.
87: The protocol handles arbitrary sized disks and views mass storage
88: as a virtual array of 512 byte blocks with no errors.
89: The lucky break for us is that third party vendors sell controllers
90: that take a SCSI disk on one side and on the other appear as a MSCP disk,
91: hiding dirty laundry such as bad block replacement and so on.
92: The end result is that we describe these controllers as regular DEC disks
93: and everything just works; at least, in theory.
94: I describe some rude exceptions below.
95: .PP
96: The software is divided into two parts.
97: The main body of code implements a file system within an arbitrary Unix file.
98: Normally, the file is a raw optical disk but it can be a regular Unix file.
99: In this latter guise, the software is much like a faster analog of
100: .I tar (1).
101: The other bunch of code is specific to SCSI and takes advantage of
102: SONY functionality beyond the basic disk functions.
103: .SH
104: The WORM software
105: .PP
106: The WORM file system code is both extensive and uninteresting.
107: It is adequately described in the manual entry.
108: The data layout on disk is a little peculiar, at least to the WORM manufacturers.
109: It is a linked list of segments, every
110: .CW "worm write"
111: command generating a new segment.
112: Each segment has the name, address,
113: .I stat
114: information, and data for a list of files.
115: The only trick here is that the pointer to the next segment
116: must be preallocated.
117: The main advantages of the design are robustness, simplicity
118: and efficiency.
119: The main disadvantage is that as more of the disk gets used,
120: it requires more CPU time to build the internal directory
121: as it is built by processing each segment in turn.
122: (This is fixed by the
123: .CW "worm btree"
124: command which build converts the directory into
125: .I cbt (1)
126: form and stores this at the end of the disk.)
127: .PP
128: The other disadvantage is that we need to recognize a block
129: that we have never written (the end of the linked list).
130: Regrettably, the WORM manufacturers regard this as an error
131: and so it is important that the MSCP-SCSI controller allows
132: the user to distinguish between this kind of error
133: (called a blank check in WORMland)
134: and a real read error (failing to read data we wrote).
135: In our environment, which admittedly deals poorly with errors of any kind,
136: this entails the driver in the operating system returning a different error
137: code (say,
138: .CW ENXIO
139: rather than the normal
140: .CW EIO ).
141: Every controller manufacturer signals blank check differently!
142: Emulex doesn't want you to know; they maintain that the mapping
143: from SCSI error codes to MSCP error codes, that is, how a blank check
144: is reported, is proprietary and won't tell you.
145: U.S. Design does exactly the right thing by returning
146: the MSCP ``forced data error'' code.
147: The MSCP standard says this is a special data error
148: whose meaning is controller-specific.
149: T.D. Systems returns a ``header compare'' data error,
150: rationalizing that no SCSI disk would ever return such an error.
151: (There is also a rumor that this is compatible with Emulex
152: but I am unable to verify.)
153: Unfortunately, other MSCP devices might so we are forced to handle
154: this special case in the MSCP driver.
155: .SH
156: The SCSI software
157: .PP
158: There are two programs:
159: .I scsish
160: and
161: .I "worm mount" .
162: The former is a shell with many commands to manipulate SCSI devices
163: although I have only tried it on SONY disks.
164: There is more than a little dependence on SONY specifics.
165: The latter does some jukebox specific stuff conveniently.
166: .PP
167: Caveat:
168: be careful about doing direct SCSI manipulation of a drive
169: you are accessing via MSCP.
170: The MSCP controllers have strong beliefs that they are the sole master of the drive
171: and are sensitive to changes in the state of the drive.
172: Unexpected changes are often rewarded by a hung controller,
173: cured only by a system reboot.
174: .PP
175: The command grammar for
176: .I scsish
177: is available via the
178: .CW help
179: command; this listing is generated from the
180: .I yacc (1)
181: grammar and is always correct.
182: Before discussing individual commands, we need to absorb some SCSI basics.
183: SCSI transactions are between an initiator and a target.
184: Both have bus ID's, a number between 0 and 7.
185: Typically, the host controller bus ID is 7 and the jukebox is 2.
186: The target is considered to have up to 8 logical units, each with a distinct
187: Logical Unit Number (LUN).
188: These numbers are typically set via DIP switches on the drives used.
189: Transactions have a SCSI status; the most common are
190: .I good
191: (the transaction succeeded),
192: .I busy
193: (the target did not respond)
194: and
195: .I check
196: (the command failed).
197: In the latter, a
198: .CW sense
199: command can be used to find out why it failed.
200: Normally,
201: .I scsish
202: reports errors rather than retrying.
203: (It will retry a few times on receiving a busy.)
204: In general, commands only take a LUN (or drive).
205: The target bus ID is set by the
206: .CW id
207: command (it defaults to the jukebox).
208: The following is a description of
209: .I some
210: of the commands.
211: An effort has been made to make the output somewhat self-explanatory;
212: if in doubt, consult the jukebox manual.
213: A basic checkout consists of
214: .CW sense~0 ,
215: .CW config ,
216: .CW status ,
217: and perhaps the diagnostics (see the
218: .CW internal
219: command).
220: Some commands accept shortened forms (like
221: .CW rel
222: for
223: .CW release ).
224: .de GG
225: .IP "\\fB\\$1\\f1" 20n
226: .if \w"\\fB\\$1\\f1"-20n \{
227: .br \}
228: ..
229: .ne 1i
230: .GG alternate
231: gives the bad block replacement numbers.
232: .GG capacity
233: gives the capacity of the disk; both the number and size of blocks
234: in decimal and hexadecimal.
235: .CS
236: capacity 0
237: drive 0: capacity 1638000x1024 (18fe70x400)
238: .CE
239: .GG config
240: describes the configuration of your jukebox, controllers,
241: and drives.
242: It gives the numbers of drives and controllers, their versions (ROM IDs)
243: and the type of Unibus controller.
244: .CS
245: config
246: config(2,0): WORM device, ' SONY WDA-3000-10 2.D', 1 controller, 2 drives
247: Unibus-SCSI controller=T.D. Systems Viking
248: ROMS: upper controller=x2, IF-129=x24, SY-46=x24, SS-30=x24
249: .CE
250: .GG copy
251: does third party copying between two drives.
252: (Third party means the initiator just says go;
253: the data only goes between the drives.)
254: I have copied an entire side of a disk in this way
255: at about 75KB/s independent of host speed.)
256: .GG disk~eject~\f4drive
257: eject the disk in the given drive.
258: .GG disk~release~\f4drive~shelf~side
259: release the disk in
260: .I drive
261: to
262: .I shelf .
263: The side (specified as
264: .CW a
265: or
266: .CW b )
267: means whether or not to invert the disk on the way to the shelf
268: (say
269: .CW b
270: to invert).
271: If the shelf and side are missing, it returns to wherever it came from.
272: .CS
273: disk rel 0 34 a
274: .CE
275: .GG disk~set~\f4shelf~side~drive
276: the opposite of
277: .CW disk~release .
278: New disks are loaded from shelf
279: .CW 127~a .
280: Note that 127 is a virtual shelf; the disks will reside on any shelf
281: which hasn't been the destination of a
282: .CW disk~release
283: command.
284: In general, the disk won't be moved to a physical drive until it needs to.
285: Thus, to ensure physical loading a disk, you need to follow a
286: .CW disk~set
287: by (say) a
288: .CW start
289: command.
290: Also, to even the load, logical units alternate between physical drives.
291: This is transparent unless one drive has a problem, in which case
292: (in a two drive system) the problems will only appear every second time
293: you use the disk.
294: .GG id~\f4number
295: set the target bus ID to
296: .I number .
297: The jukebox, and default, is 2.
298: .GG inquiry~\f4drive
299: give a simple status for the the given drive or if no drive is given, all drives.
300: .CS
301: inq
302: drive 2,0: disk,write protect,,,ready (0x9)
303: drive 2,1: disk,writable,,,not ready (0x0)
304: drive 2,2: no disk,writable,,,not ready (0x40)
305: drive 2,3: no disk,writable,,,not ready (0x40)
306: drive 2,4: no disk,writable,,,not ready (0x40)
307: drive 2,5: no disk,writable,,,not ready (0x40)
308: drive 2,6: no disk,writable,,,not ready (0x40)
309: drive 2,7: no disk,writable,,,not ready (0x40)
310: .CE
311: .GG internal~\f4n
312: there are several internal diagnostics; the full list is given by
313: .CS
314: internal
315: internal 0: internal command table
316: internal 1: error information table
317: internal 2: arm controller diagnostics
318: internal 3: scsi control board diagnostics
319: internal 4: drive controller diagnostics
320: internal 5: jukebox status
321: .CE
322: The arm controller diagnostics take about 4 minutes,
323: the drive controller diagnostics take about 100 seconds.
324: The others are quite short.
325: The drive controller diagnostic needs a LUN.
326: .CS
327: internal 3
328: scsi control board diagnostics:
329: ended normally (time: 7s)
330: internal 4 0
331: drive 0[upper]: drive controller diagnostics
332: diagnostic result: no faults (time: 97s)
333: test 0[drive on/off]: good
334: test 1[read disk id]: good
335: test 2[move]: good
336: test 3[seek]: good
337: test 4[blank sector search]: good
338: test 5[written sector search]: good
339: test 6[search writable area]: good
340: test 7[write]: diagnostic could not be done
341: test 8[ECC margin check]: diagnostic could not be done
342: test 9[read data compare]: diagnostic could not be done
343: diagnostic count (drive:avail): 0:199 1:0 2:0 3:0 4:0 5:0 6:0 7:0
344: internal 5
345: jukebox status: component(fatal err/err/cmds)
346: upper drive(0/0/38) lower drive(0/0/65) sys control(0/0/115) backup mem(0/4/4454)
347: .CE
348: The
349: .CW internal~4
350: diagnostic requires a LUN with a disk in it.
351: .GG media~\f4drive~start~nblocks
352: do a media reliability check for the given blocks
353: (it does about 200 blocks/sec).
354: Prefixing the media command with
355: .CW ext
356: gives details for every funny block.
357: You can also dump the information to a disk file.
358: .CS
359: media 1 0 1000
360: drive 1: media check for 1000 blocks [0-999], lower drive
361: 999 good, 1 unwritten,
362: ext media 1 0 1000
363: drive 1: detailed media check for 1000 blocks [0-999], lower drive
364: lbn 0: unwritten
365: 999 good, 1 unwritten,
366: .CE
367: .GG read~id~\f4drive
368: gives the id as used by the
369: .I worm
370: commands.
371: .CS
372: read id 1
373: id='backupdb2a'
374: .CE
375: .GG reset
376: resets the controller.
377: The behavior depends on the controller type.
378: The T.D. Systems controller resets quickly and reliably.
379: The U.S. Design controller causes a SCSI bus reset which resets the jukebox.
380: This takes about 90 seconds.
381: .GG sense~\f4drive
382: gives the sense data for the last operation.
383: A prefix of
384: .CW ext
385: gives the extended sense.
386: .CS
387: sense 0
388: sa=0x0 mscp=0x0 status=0x0(good) (ext: no sense)
389: 70 0 0 0 dd dd dd dd dd dd dd dd dd dd dd dd
390: sense 1
391: sa=0x0 mscp=0x0 status=0x0(good) (ext: illegal request)
392: 70 0 5 0 dd dd dd dd dd dd dd dd dd dd dd dd
393: .CE
394: .GG sleep~\f4number
395: sleep for
396: .I number
397: seconds.
398: This gives you time to rush to the hardware!
399: .GG start~\f4drive
400: start the given drive.
401: This is necessary before most commands for a disk.
402: .GG status
403: tells you all sorts of stuff, some of it interesting.
404: .CS
405: status
406: drive 0: ready,disk in LUN,power on,disk in drive 0, return shelf 18a (36)
407: drive 1: not ready,no disk in LUN,power on,disk in shelf 0
408: drive 2: not ready,no disk in LUN,power on,disk in shelf 0
409: drive 3: not ready,no disk in LUN,power on,disk in shelf 0
410: drive 4: not ready,no disk in LUN,power on,disk in shelf 0
411: drive 5: not ready,no disk in LUN,power on,disk in shelf 0
412: drive 6: not ready,no disk in LUN,power on,disk in shelf 0
413: drive 7: not ready,no disk in LUN,power on,disk in shelf 0
414: 0: disk,
415: 5: disk,
416: 6: disk,
417: 8: disk,
418: 18: no disk
419: 21: disk,
420: I/O shelf: no disk
421: carrier: disk shelf=8a (16)
422: upper drive: disk, LUN=0
423: lower drive: no disk
424: .CE
425: .GG stop~\f4drive
426: stop the drive from spinning.
427: .GG test~\f4drive
428: does a Test Unit Ready.
429: It simply returns a good/check answer.
430: .SH
431: Reliability
432: .PP
433: My WORM experience covers four pieces of hardware:
434: WORM drive, WORM controller, jukebox and the UNIBUS-SCSI controller.
435: .PP
436: The WORM drives have proved fairly reliable.
437: Two of the six drives in our installation have required service.
438: One has an intermittent loading problem where the disk being loaded
439: is continually inserted and ejected.
440: The other was a brand new drive in the jukebox.
441: It gave no overt signs of trouble but media quality diagnostics
442: done in that drive always gave terrible results.
443: The drives in the jukebox can be used by themselves;
444: the only difference is a DIP switch setting
445: (which regrettably requires taking off the drive's front panel).
446: .PP
447: The WORM controllers had an early problem with infant mortality
448: (two of four were D.O.A.).
449: There has been no trouble since then.
450: .PP
451: The jukebox really has two parts.
452: One is the robotics section and the other is a smart controller
453: which makes the jukebox look like 8 logical drives
454: independent of how many physical drives there are.
455: (In fact, SONY measures the time to change disks by mounting two
456: disks in a one drive jukebox and then copying from one to another.)
457: The controller has worked flawlessly.
458: In fact, be warned;
459: it has battery backup and remembers where disks were and are!
460: The mechanical parts have been pounded on and so far have not had problems.
461: All in all, our jukebox experience is limited but positive.
462: The maintenance contract on the jukebox is relatively inexpensive.
463: .PP
464: The UNIBUS-SCSI controller has been the most exasperating piece of the puzzle.
465: The first attempt was the Emulex UC13 controller.
466: It never really worked quite right and Emulex
467: refused to say how a blank check was returned.
468: The second attempt was the U.S. Design 1158.
469: After four or five months,
470: they were able to decide exactly what the switch settings
471: were supposed to be and the MSCP emulation worked just fine.
472: The SCSI controller (that is, an 1158 acting as a pure SCSI controller)
473: seems to work okay (apart from not working in single user mode).
474: Unfortunately, the combination of the two controllers
475: (necessary for controlling the jukebox) is still flaky.
476: The MSCP controller has a tendency to hang and the SCSI controller
477: generates stray interrupts.
478: .PP
479: The third and most successful controller has been the T.D. Systems
480: Viking UDD controller.
481: This combines both controllers on one quad high board
482: (as opposed to two hex high boards for the U.S. Design setup)
483: for the same price as one U.S. Design board.
484: Despite some unpleasantness from the manufacturer (or more exactly, its president),
485: these boards have worked and worked well.
486: .SH
487: Ordering information
488: .PP
489: The SONY WORMS and my software are deployed at three sites within Area 112.
490: The distinct configurations are listed below;
491: the total number of WORM drives is 9.
492: .TS
493: center;
494: c c c
495: c a l.
496: Site Machine Configuration
497: _
498: 1121 8550 1 controller/1 drive
499: 1122 11/750 1 controller/1 drive
500: 1122 11/750 jukebox (1 controller/1 drive)
501: 1127 11/750 1 controller/2 drives
502: 1127 11/750 jukebox (1 controller/2 drives)
503: .TE
504: .PP
505: The WORM hardware was purchased from U.S. Design
506: because they offered the UNIBUS controller.
507: If you want to buy from someone closer than Maryland, call SONY for a list
508: of distributers.
509: All costs are approximately correct as of mid-1988.
510: .TS
511: center;
512: c c c c
513: l a l.
514: Part Part # Cost Supplier
515: _
516: SONY disk controller WDA-3000 controller $10K U.S. Design
517: SONY disk drive WDA-3000 disk drive $12.5K U.S. Design
518: SONY jukebox WDA-3000-10 Auto Changer $75K U.S. Design
519: 1158 controller B58V-3B $1895 U.S. Design
520: T.D. Systems UDD controller Viking UDD controller $1800 Trimarchi
521: .TE
522: You will need a converter to connect the 50 pin Berg connector on the
523: controller boards to the 50 pin D connector on the SONY gear.
524: We use ones that came with the U.S. Design controller.
525: It is not hard to make your own.
526: .TS
527: center;
528: c s
529: a l.
530: Phone numbers
531: _
532: (201) 930-6030 SONY (Bart Connors)
533: (617) 937-9465 T.D. Systems
534: (800) 356-6638 Trimarchi Inc (distributor for T.D. Systems in NJ)
535: (301) 577-2880 U.S. Design (Harry Garonzik)
536: .TE
537: .SG
538: .sp 2
539: Att.
540: .br
541: manual pages
542: .bp
543: worm(1)
544: scsi(4)
This archive runs on limited infrastructure. Preserving old code on modern bandwidth. Automated agents are requested to crawl responsibly.