Description:

Large LUNs on the SE3310 have problems building a file system with newfs.

This is an example of what kind of problems you may run into:

Note the cylinder count below is 43776:

# prtvtoc /dev/rdsk/c3t0d0s2

* /dev/rdsk/c3t0d0s2 partition map

*

* Dimensions:

* 512 bytes/sector

* 127 sectors/track

* 127 tracks/cylinder

* 16129 sectors/cylinder

* 43776 cylinders

* 43774 accessible cylinders

*

* Flags:

* 1: unmountable

* 10: read-only

*

* First Sector Last

* Partition Tag Flags Sector Count Sector Mount Directory

0 2 00 0 274193 274192

1 3 01 274193 274193 548385

2 5 01 0 706030846 706030845

6 4 00 548386 705482460 706030845

Example of 'iostat -En' output (see the LUN is approximately 360GB):

c3t0d0 Soft Errors: 0 Hard Errors: 0 Transport Errors: 0

Vendor: SUN Product: StorEdge 3310 Revision: 0325 Serial No:

00000017985D0A00

Size: 361.50GB <361496051200 bytes>

Media Error: 0 Device Not Ready: 0 No Device: 0 Recoverable: 0

Illegal Request: 0 Predictive Failure Analysis: 0

This is typically what you'll see if it's a filesystem 200GB or larger:

# newfs /dev/rdsk/c3t0d0s6

newfs: construct a new file system /dev/rdsk/c3t0d0s6: (y/n)? y

With 16129 sectors per cylinder, minimum cylinders per group is 16

This requires the fragment size to be changed from 1024 to 4096

Please re-run mkfs with corrected parameters

This newfs example actually completes, but also complains:

# newfs -f 4096 /dev/rdsk/c3t0d0s6

newfs: construct a new file system /dev/rdsk/c3t0d0s6: (y/n)? y

Cylinder groups must have a multiple of 16 cylinders with the given

parameters

Rounded cgsize up to 256

Warning: insufficient space in super block for

rotational layout tables with nsect 127, ntrack 127, and nrpos 8.

Omitting tables - file system performance may be impaired.

Warning: 4 sector(s) in last cylinder unallocated

/dev/rdsk/c3t0d0s6: 705482456 sectors in 43740 cylinders of 127 tracks,

127 sectors

344473.8MB in 2734 cyl groups (16 c/g, 126.01MB/g, 15808 i/g)

super-block backups (for fsck -F ufs -o b=#) at:

32, 258224, 516416, 774608, 1032800, 1290992, 1549184, 1807376, 2065568,

2323760, 2581952, 2840144, 3098336, 3356528, 3614720, 3872912, 4131104,

4389296, 4647488, 4905680, 5163872, 5422064, 5680256, 5938448, 6196640,

6454832, 6713024, 6971216, 7229408, 7487600, 7745792, 8003984, 8262176,

8520368, 8778560, 9036752, 9294944, 9553136, 9811328, 10069520, 10327712,

10585904, 10844096, 11102288, 11360480, 11618672, 11876864, 12135056,

12393248, 12651440, 12909632, 13167824, 13426016, 13684208, 13942400,

14200592, 14458784, 14716976, 14975168, 15233360, 15491552, 15749744,

16007936, 16266128, 16524320, 16782512, 17040704, 17298896, 17557088,

17815280, 18073472, 18331664, 18589856, 18848048, 19106240, 19364432,

19622624, 19880816, 20139008, 20397200, 20655392, 20913584, 21171776,

21429968, 21688160, 21946352, 22204544, 22462736, 22720928, 22979120,

23237312, 23495504, 23753696, 24011888, 24270080, 24528272, 24786464,

25044656, 25302848, 25561040, 25819232, 26077424, 26335616, 26593808,

26852000, 27110192, 27368384, 27626576, 27884768, 28142960, 28401152,

28659344, 28917536, 29175728, 29433920, 29692112, 29950304, 30208496,

30466688, 30724880, 30983072, 31241264, 31499456, 31757648, 32015840,

.......

Solution section

Note: (The following solutions were obtained from http://www.dothill.com/support. From there you can go to the SANsolve Knowledge Base and search for these Issues.)

Solution number 1:

Issue 1050 from SanSolve (Dot Hill's solution site):

Solaris, Head, Cylnder and Sector capacity

relabel issue

Product:

SANnet General

Description:

On Solaris, there is a capacity issue you have

to be aware of when reaching 200 - 250gb. All

Solaris versions 2.5 thru 2.8.

Solution:

In the Raid controller

Main Menu:

View and Edit Configuration Parameters

Host-side SCSI Parameters

Host Cylinder/Head/Sector Mapping Configuration

Sector Ranges - 255 Sectors

Head Ranges - 64 Heads

Cylinder Ranges - < 65536 Cylinders

Once this is adjusted there is no need of a

reset of the controller settings are immediate,

nor rebooting your server till everything is

re-labeled.

(note)Solaris will need to re-label the disk in

order for the new parameters of the disk to be

recognized. (this is described below)

As root, type:

Format [enter]

from the "AVAILABLE DISK SELECTIONS" select your

Sannet disk from the menu and [enter].

then type:

type [enter]

you will then select 0 for Autoconfigure. (this

will re-read the disk parameters in question)

when finished you will want to type:

l [enter] (this is for the label command, this

will write and save the new label to the disk,

if this is not done, your autoconfig will not

take effect.)

then type:

q [enter] this will exit you from the format

menu. You will then be able to write a

filesystem to your disk with the command newfs

/dev/rdsk/c#t#d#s#.

Solution number 2:

Issue 78:

Solaris does not let me newfs large Logical

Drives

Product:

SANnet General

Description:

Solaris does not let me newfs large Logical

Drives

Solution:

There is a limit for Cylinders under Solaris.

If you have a large Logical Drive (Larger than

512GB), perform the following steps from the

Raid Controller RS-232 connection:

1. Select "View and Edit Configuration

Parameters"

2. Select "Host side SCSI Parameters"

3. Select "Host Cylinder/Head/Sector Mapping

Configuratioin"

4. Modify the Head Ranges to - 64 Heads

5. Modify the Cylinder Ranges to - < 32768

Cylinders

You may also have to do the following under

Solaris:

1. Format

2. Select the desired drive

3. type

4. Auto Configure

5. label the drive

Internal Section