JumpStart配置客户机的RAID-1时的metadb疑惑？

我遇到和下面这个仁兄一模一样的问题，也是在x86平台。

Hi,
  I'm trying to create a mirrored / and swap during the jumpstart build process, but I keep getting errors when JS tries to create the metadbs.  I'm currently trying Nevada b52 on x86 (on VMWare with 2 emulated SCSI disks), but also tried S10 0606 with the same results.

My jumpstart profile has:
install_type    initial_install
system_type     standalone
partitioning    explicit
fdisk           all solaris maxfree
metadb          c1t0d0s3---------默认会创建3个metadb，每个8192块，共24576块
metadb          c1t1d0s3
filesys         mirror c1t0d0s0 c1t1d0s0 7300 /
filesys         mirror c1t0d0s1 c1t1d0s1 500  swap

During the JS install, the process errors out with
---------------------------------------------------------
Creating SVM State Replica on disk c1t0d0s3
metadb: test2: c1t0d0s3: device size 24576 is too small for metadevice database replica

Creating SVM State Replica on disk c1t1d0s3
metadb: test2: c1t1d0s3: device size 24576 is too small for metadevice database replica
---------------------------------------------------------

If I look at the JS installer's /tmp/create_mirror script, I see
....
/usr/bin/echo "Creating SVM State Replica on disk c1t0d0s3"
/usr/sbin/metadb -a -f -c 3 -l 8192 c1t0d0s3
/usr/bin/echo "Creating SVM State Replica on disk c1t1d0s3"
/usr/sbin/metadb -a -f -c 3 -l 8192 c1t1d0s3
....

The disks appear to be partitioned fine (both have the settings below)
---------------------------------------------------------
Part      Tag    Flag     Cylinders        Size            Blocks
  0       root    wm     251 - 3900        7.13GB    (3650/0/0) 14950400
  1       swap    wu       1 -  250      500.00MB    (250/0/0)   1024000
  2     backup    wm       0 - 4091        7.99GB    (4092/0/0) 16760832
  3 unassigned    wm    3901 - 3906       12.00MB    (6/0/0)       24576----jumpstart自动把分片3分配了24576个块
  8       boot    wu       0 -    0        2.00MB    (1/0/0)        4096
---------------------------------------------------------

If I manually run the metadb command, one thing I noticed is that the first metadb is placed at an offset of 16 blocks (I assume to prevent over-writing the bootblock).  What I think may be the problem is that JS is allocating only exactly enough space on the slice for the metadbs, but really it should be allocating that space +8k to allow for the 16 block offset.

CODE:

在jumpstart过程中断后，进入交互式shell。如果我手工创建：
# metadb -af -c 2 /dev/dsk/c1t0d0s3
# metadb
显示第一个metadb从第16块开始。。。我怀疑是这个原因，导致第三个metadb无法创建。因为总共只有24576块，给它偏移了16块后，留给第三个块的空间不够8192块了。。。难道是X86的bug？

searching......



相关阅读

Solaris的NFS使用手册

Solaris基础和常用知识（新手必读）

Solaris 操作系统的最新UNIX基本知识官方教材中文版


以下这些博文,你可能更感兴趣：

---