［付録］ Apollo DOMAIN_OS の使い方３


● Apollo DOMAIN 久しぶりの作業   2013/11/s

   * ロ−カルレジストリのデ−タをバックアップ

    $ wd //node_aaaaa/sys/registry
    $ cpf rgy_local /nix/cad/temp
    $ cpf rgy_local.bak /nix/cad/temp

    node_aaaaa は親機、node_bbbbb はディスクレス機。

   * 

    $ lcnode -me
    The node ID of this node is BBBBB.

    bbbbb で root でログイン。ロ−カルレジストリは使える。

    aaaaa に $ crp -on //node_aaaa -me して。root のまま。


    $ lusr
        root.staff.none.AAAAA           //node_aaaaa

    $ /etc/edrgy -l
    Using local registry.
    edrgy=> v
    root[.staff.none]:w0UuY0sQV7sAI:0:10::/::
    ktest[.none.none]::300:12::/nix/cad/ktest::
    edrgy=> q

    $ login
    login: root
    Password: 
    Using local registry. Can't use network registry:  - Registry server unavailable (from RGYC / Server)
    Account entry has expired or is marked invalid.
    username[.group[.org]]
    login:


    動いているデ−モンが何か足らないのでないか。glbd が動いてなかったのか、
    それともいろいろいじっている内に自分で glbd を止めた気がする。
    $ /etc/server /etc/ncs/glbd & やっても $ /etc/ncs/glbd & やっても、glbd
    は起動してこなかった。このマシン、親機を電源入れ直して再起動してみない
    といけないな。


    $ date
    Tuesday, November 6, 2012   11:06:11 am (JST)

    $ ld /sys/node_data/etc/daemons                          
    Directory "/sys/node_data/etc/daemons":

    README  cron    glbd    llbd    rgyd


    $ lcnode
     The node ID of this node is AAAAA.
     1 other node responded.

    Node ID      Boot time           Current time         Entry Directory

    AAAAA   2012/06/21  1:58:38   2012/11/06 11:15:15  //node_aaaaa
    BBBBB   2008/05/10  9:32:44   2008/05/18 11:35:19  *** DISKLESS ***  partner node: AAAAA


    $ pst
    ------------------------------------------------------------
    Node:  AAAAA
    Time:  Tuesday, November 6, 2012   11:07:16 am (JST)
    ------------------------------------------------------------
     Processor | PRIORITY |  Program | State |   Process Name
     Time (sec)| mn/cu/mx |  Counter |       |
    ------------------------------------------------------------
    11849573.217  -- -- --   --------   -----   <Null Process>
      20663.909  -- -- --   --------   -----   <OS Processes>
         12.281  16/16/16   7608A494    Wait   init
          0.917   3/14/14   76089DB4    Wait   mbx_helper
      11109.933   3/ 7/ 7   7608A36E    Wait   rgyd.1.3
      29801.864   3/14/14   7608A36E    Wait   uid = E8FC2654.F00AAAAA (rpcd)
          0.436   3/14/14   7608A494    Wait   root.BBBBB
       2086.552   3/13/14   7608A36E    Wait   uid = E8FC2785.B00AAAAA (cron)
          1.870  16/16/16   7608A36E    Wait   display_manager
          0.402   3/14/14   7608A36E    Wait   uid = E8FC2789.000AAAAA (lpd)
          0.064   3/14/14   <active>   Ready   uid = EBB415DC.C00AAAAA (pst)
        141.917   3/14/14   7608A36E    Wait   server_process_manager
         27.092   3/14/14   76089DB4    Wait   netman
      25347.849   3/14/14   7608A36E    Wait   uid = E8FC2796.900AAAAA (prmgr)
      15189.192   3/13/14   7608A36E    Wait   prsvr:p:0
    -----------
    11953957.894


    $ ld /sys/registry
    rgy_data       rgy_local      rgy_local.bak


    $ ld -a /sys/registry
    sys   type      blocks  current
    type  uid         used   length   attr rights       name

    dir   nil            4      4096  P    prwx-        rgy_data
    file  nil           28     26586  P    prwx-        rgy_local
    file  nil           28     26586  P    prwx-        rgy_local.bak


    $ ld -a /sys/registry/rgy_data
    sys   type      blocks  current
    type  uid         used   length   attr rights       name

    file  unstruct      12      8204  P    prwx-        acct
    file  unstruct       8      4628  P    prwx-        group
    file  unstruct       4      1528  P    prwx-        org
    file  unstruct       8      7452  P    prwx-        person
    file  unstruct       4       152  P    prwx-        replicas
    file  unstruct       4       148  P    prwx-        rgy
    file  unstruct       4        74  P    prwx-        rgy_state
    dir   nil            4      4096  P    prwx-        unix
    file  unstruct       4      1024  P    prwx-        update_log

    $ ld -a /sys/registry/rgy_data/unix
    sys   type      blocks  current
    type  uid         used   length   attr rights       name

    file  unstruct       4       206  P    prwx-        group
    file  unstruct       4       205  P    prwx-        org
    file  unstruct       4       865  P    prwx-        passwd


    $ catf org
    wheel::0:
    apollo::1:user
    none::12:root,daemon,none,user,lp,admin,sys_person,uucp,bin,\
        ktest
    sys_org::13:


    $ catf group
    wheel::0:
    daemon::1:
    bin::3:root
    mail::6:bin
    staff::10:
    none::12:alias,qmaild
    sys_proj::13:
    locksmith::14:
    login::15:
    backup::16:user
    sys_admin::17:user
    server::18:
    sys::19:root
    qmail::2107:
    nofiles::2108:

    $ catf passwd                   $ catf /etc/passwd  もこれを見ているみたい。
    root:w0UuY0sQV7sAI:0:10::/:
    daemon:w0UuY0sQV7sAI:1:12::/:
    bin:w0UuY0sQV7sAI:3:3::/:
    uucp:w0UuY0sQV7sAI:4:12::/usr/spool/uucppublic:/usr/lib/uucp/uucico
    none:w0UuY0sQV7sAI:12:12::/:
    sys_person:w0UuY0sQV7sAI:13:12::/:
    user:w0UuY0sQV7sAI:14:12::/:
    admin:w0UuY0sQV7sAI:15:12::/:
    lp:w0UuY0sQV7sAI:16:3::/:
    ktest::300:12::/nix/cad/ktest:


    $ /etc/edrgy 
    ?(edrgy) Warning: the registry at site "dds://node_aaaaa" is read-only.
    edrgy=> q


    $ /etc/edrgy -s //node_aaaaa
    ?(edrgy) Warning: the registry at site "//node_aaaaa" is read-only.
    edrgy=> q

    レジストリ−サ−バの手動起動
    $ /etc/server -p /etc/rgyd &

    $ /etc/ncs/glbd &   glbd と rgyd をいったん殺して、再度稼働させる場合。
    $ /etc/rgyd         glbd はすぐ稼働する。rgyd は 数十秒かかる、& は付けないように。

    この場合スレ−ブレジストリデ−モンを sigp xxx -b で止めると管理が
    マスタ−側に戻る。


    $ /etc/edrgy -l
    Using local registry.
    edrgy=> v
    root[.staff.none]:w0UuY0sQV7sAI:0:10::/::
    ktest[.none.none]::300:12::/nix/cad/ktest::
    edrgy=> do p
    Domain changed to: person
    edrgy=> v
    root             0      ktest          300  
    edrgy=> do a
    Domain changed to: account
    edrgy=> c
    change account=> Enter account id [p.g.o]: ktest.none.none
    Enter new account id [p.g.o]: (cannot be changed)
    Enter new abbreviation type [p/pg/pgo]: (p) 
    Enter new password: (cannot be changed)
    Enter new misc info in quotes: () 
    Enter new home directory: (/nix/cad/ktest) 
    Enter new shell: () (cannot be changed)
    Password valid [y/n]? (y) (cannot be changed)
    Enter new expiration date [yy/mm/dd or 'none']: (2008/03/20) (cannot be changed)
    Account valid for login [y/n]? (y) n
    change account=> Enter account id [p.g.o]: 
    edrgy=> q


   * rgyd 起動しなおした  2013/11/s

    $ sigp rgyd.1.3 -s      止まった。

    $ pst
    ------------------------------------------------------------
    Node:  AAAAA
    Time:  Tuesday, November 6, 2012   1:12:41 pm (JST)
    ------------------------------------------------------------
     Processor | PRIORITY |  Program | State |   Process Name
     Time (sec)| mn/cu/mx |  Counter |       |
    ------------------------------------------------------------
    11857015.087  -- -- --   --------   -----   <Null Process>
      20685.641  -- -- --   --------   -----   <OS Processes>
         12.281  16/16/16   7608A494    Wait   init
          0.999   3/14/14   76089DB4    Wait   mbx_helper
          0.215   3/14/14   7608A36E    Wait   uid = EBB4859D.900AAAAA (rgyd)
      29821.374   3/14/14   7608A36E    Wait   uid = E8FC2654.F00AAAAA (rpcd)
          0.909   3/14/14   7608A494    Wait   root.BBBBB
       2087.921   3/14/14   7608A36E    Wait   uid = E8FC2785.B00AAAAA (cron)
          1.870  16/16/16   7608A36E    Wait   display_manager
          0.402   3/14/14   7608A36E    Wait   uid = E8FC2789.000AAAAA (lpd)
          0.653   3/ 7/ 7   7608A36E    Wait   rgyd.1.3
          0.064   3/14/14   <active>   Ready   uid = EBB485FD.800AAAAA (pst)
        142.017   3/14/14   7608A36E    Wait   server_process_manager
         27.095   3/14/14   76089DB4    Wait   netman
      25363.671   3/14/14   7608A36E    Wait   uid = E8FC2796.900AAAAA (prmgr)
      15198.717   3/14/14   7608A36E    Wait   prsvr:p:0
    -----------
    11950359.184


    しばらくしたら uid = EBB4859D.900AAAAA (rgyd) は rgyd.1.3 になっていた。

    だめ、katou でも root でもログインできない。ロ−カルのデ−タベ−スも
    みないというか、アカウントの有効期限がきれているとなってしまう。


● rgyd

   * rgyd

    $ khelp rgyd

    オプション
     -create     ネットワーク・レジストリのレプリカを作成します。このオプションによ
                 って，レジストリ・データベースのコピーが作成され, スレーブ・サーバ
                 ・プロセスが開始されます。-createオプションは，あるノードでスレー
                 ブ・サーバ・プロセスを初めて開始する場合に限って使用してください。
                 デーモンを再開始する場合には，オプションは何も指定する必要がありま
                 せん。マスタ・レプリカを作成するには，cvtrgy（SR9のレジストリをSR10
                 のフォーマットに変換する場合）または rgy_create（新しいSR10レジスト
                 リを作成する場合）を使用してください。

     -recreate   スレーブ・レプリカを作成し直します。このオプションは，データベース
                 のスレーブのコピーが修復不可能な状態に破壊されてしまった場合に限っ
                 て使用してください。既存のデータベースが破棄され，新しいものが作成
                 されます。

     1. rgy_create または cvtrgy を使用してマスタ・データベースを作成した後，
        レジストリのマスタ・レプリカを開始します。
        $ /etc/server -p /etc/rgyd

     2. レジストリのスレーブ・レプリカを開始します。
        $ /etc/server -p /etc/rgyd -create

     3. レジストリの既存のレプリカ(マスタあるいはスレーブ)を再開始します。
        $ /etc/server -p /etc/rgyd

     4. リモート・ホスト //yak のレジストリの既存のレプリカを再開始します。
        $ crp -on //yak -cps //yak/etc/rgyd


   * glbd

     $ khelp glbd

     glbd の開始時には，ローカル・ノード上でローカル・ロケーション・ブローカ
     ・デーモン（llbd）が実行中でなければなりません。通常は，どちらの
     ブローカもブート時に /etc/rcファイルから始動されます。

     glbd が IPプロトコルを介して通信を行う場合には，ローカル・ノード上で
     TCPデーモン（tcpd）も実行中でなければなりません。tcpd は llbd よりも
     前に始動している必要があります。

     1. このネットワークまたはインタネット上の GLB の最初のレプリカを
        生成し，開始します。
     $ /etc/server /etc/ncs/glbd -create -first -family dds &

     2. 後続の GLB のレプリカを始動し，そのデータベースをホスト //jeeves
        から初期化します。
     $ /etc/server /etc/ncs/glbd -create -from dds://jeeves &

     3. GLB の既存のレプリカを再始動します。
     $ /etc/server /etc/ncs/glbd &

     4. リモート・ホスト //bertie上の GLB の既存のレプリカを再始動します。
     $ crp -on //bertie /etc/server //bertie/etc/ncs/glbd &


   * glbd と rgyd を初期化して、マスタ−とする。

    $ /etc/server /etc/ncs/glbd -create -first &
    $ /install/tools/rgy_create

    $ /etc/ncs/glbd &   glbd と rgyd をいったん殺して、再度稼働させる場合。

    $ ld /etc/daemons   とりあえずこれだけ空ファイルを作り、リブ−トする。
    inetd    tcpd   glbd  llbd  rgyd


●

   * `2g/08/s のこと

    node_bbbbb から node_aaaaa に crp で入って。

    $ /etc/rgy_admin
        Default object: rgy  default host: dds://node_aaaaa
        State: in service  slave
    rgy_admin: ?
    Known commands are:
        set             site            i[nfo]          stop
        init[rep]       repr[ep]        delr[ep]        lr[ep]
        mon[itor]       reset           ch[ange_master] become
        state           q[uit]          h[elp]          ? 


    node_bbbbb にて。

    $ /etc/rgy_admin
        Default object: rgy  default host: unspecified
    rgy_admin: q
    $


   * 2008/05/s のこと

    $ /etc/rgy_admin
    	Default object: rgy  default host: dds://node_aaaaa
    	State: in service  slave

    rgy_admin: mon
    (master)  dds:#0.ccccc   	*** not communicating ***	
              dds://node_aaaaa	state: in service        2006/06/26.13:40:03	
              ---

    rgy_admin: info
    	Default object: rgy  default host: dds://node_aaaaa
    	State: in service  slave


● 

    /sys/node_data.bbbbb/etc/rc.local
    -----------------------------------------------------------
    |if [ -f /etc/ifconfig ]; then
    |   /etc/ifconfig net0 `/etc/hostname` netmask defaultmask
    |   /etc/ifconfig eth0 192.168.1.9
    |   /etc/ifconfig lo0  127.0.0.1       
    |fi
    |
    |/etc/route add net 0.0.0.0 192.168.1.1 1
    
    $ ld /etc/daemons/
    Directory "/sys/node_data.bbbbb/etc/daemons":
    README   Xapollo  tcpd


● 

   * 2010/09/m

    Apollo モニタが画面が小さくさざ波を打っている。落ち着かせたいのだが調整できない。
    まあそこそこみれるようになったので、まあいいか。


   * 2006/06/m 

    $ ld //node_ccccc/sys/node_data/etc/daemons
    README      Xapollo     automount   cron        glbd        inetd
    llbd        lpd         mountd      nfsd        portmap     rgyd
    rpc.lockd   rpc.mountd  rpc.statd   tcpd        xpager2


● kbs_patch.c

/*
$Log:	kbs_patch.c,v $
 * Revision 1.2  99/01/23  19:57:29  paul
 * Correct Version macro.
 * Minor error message tweak.
 * 
 * Revision 1.1  99/01/23  18:31:56  paul
 * Initial revision

    This program fixes a limitation in the Domain/OS kernel that
prevents SCSI drives with a capacity greater than 4GB (4*1024**3) from
being INVOL'd.  When attempting to INVOL such a drive, the reported
capacity is reported as 4GB less than the true capacity.  For example, a
4,569,599,488 byte drive was reported as having a capacity of
approximately 268MB.
    The problem is that the kernel code that determines the capacity of
a physical volume is not written for capacities that exceed 2^32 bytes.
The OS determines how many blocks (1K or 4K) the drive can support by
issueing the ReadCapacity SCSI command.  This command returns the number
of blocks on the drive, and the size of a block.  Then the system
multiplies these numbers together, which yields the capacity of the
drive, in bytes.  For the last step, the system divides the byte
capacity by the system page size (1K or 4K).  However, a byte capacity
that is greater than 2^32-1 will have overflowed, and the result will be
incorrect.  One example if a drive with a reported capacity of 8,924,999
blocks, at 512 bytes per block, which comes to 4,569,599,488 total
bytes (0x1105e8e00).  This number exceeds 32 bits, and becomes
0x105e8e00, which divides by the page size of the node to become either
0x105e8 or 0x417a3 blocks, depending on the page size.
    Here is a code fragment for the capacity calculation:
        MOVE.l      (a2),d0            * d0 = numblocks
        MULS.l      ($4,a2),d0         * d0 *= blocksize
        MOVE.l      d0,d1              * d1=d0
        LSR.l       #$8,d1             * d1 /= 4096 (page size)
        LSR.l       #$4,d1             *
        MOVE.l      d1,($FFFFFFE0,a6)  * local_b_per_pvol = d1
On a machine with a 1K pagesize, the divisor is 1K, and the second shift
instruction has a count of 2.
    The simplest fix for the problem is to reorder the code that does
the multiply/divide.  Note that the division is done by two adjacent LSR
instructions.  If one of these is moved to operate on the blocksize,
then the multiply won't overflow.  Here is an example:
        MOVE.l      ($4,a2),d0         * d0 = blocksize
        LSR.l       #$8,d0             * d0 /= 256
        MULS.l      (a2),d0            * d0 *= numblocks
        MOVE.l      d0,d1              * d1=d0
        LSR.l       #$4,d1             * d1 /= 16
        MOVE.l      d1,($FFFFFFE0,a6)  * local_b_per_pvol = d1
This solution worked in the case of the 4.5G drive mentioned earlier.

Notes:
    This solution depends on the drive's block size being a multiple of
256.  If the block size does not meet this requirement, then the results
will be incorrect.
    The patched kernel is only necessary when the drive is being
INVOL'd.  Once invol has completed, the drive may be mounted on a normal
system.
*/

#systype "sys5.3"

#include <stdio.h>
#include <string.h>
#include <stdlib.h>
#include <sys/types.h>
#include <sys/stat.h>

#include <apollo/base.h>
#include <apollo/error.h>
#include <apollo/ms.h>

/* prototype for undocumented function */
unsigned long getpagesize(void);

/* version */
#define Version (strtok("$Revision: 1.2 $"+10, " "))

/* search/replace information for machines with a 1k page size */
unsigned short search_pattern1k[] = {0x2012, 0x4C2A, 0x0000, 0x0004, 0x2200, 0xE089, 0xE489, 0x2D41};
#define search_pattern_length1k (sizeof(search_pattern1k)/sizeof(search_pattern1k[0]))
unsigned short replace_pattern1k[] = {0x202a, 0x0004, 0xe088, 0x4c12, 0x0000, 0x2200};
#define replace_pattern_length1k (sizeof(replace_pattern1k)/sizeof(replace_pattern1k[0]))

/* search/replace information for machines with a 4K page size */
unsigned short search_pattern4k[] = {0x2012, 0x4C2A, 0x0000, 0x0004, 0x2200, 0xE089, 0xE889, 0x2D41};
#define search_pattern_length4k (sizeof(search_pattern4k)/sizeof(search_pattern4k[0]))
unsigned short replace_pattern4k[] = {0x202a, 0x0004, 0xe088, 0x4c12, 0x0000, 0x2200};
#define replace_pattern_length4k (sizeof(replace_pattern4k)/sizeof(replace_pattern4k[0]))


/* Search for the pattern, return VA of the pattern, or NULL if not found. */
unsigned short *search(unsigned short *base, unsigned long Words, unsigned short *search_pattern, unsigned long search_length)
   {
   int count;
   unsigned short *retval;

   /* Adjust Words for pattern length */
   Words -= (search_length-1);
   retval = NULL;
   while (Words--)
      {
      /* See if we match at the current location */
      for (count = 0; count < search_length; count++)
         if (base[count] != search_pattern[count])
            break;
      if (count == search_length)
         {
         if (retval)
            {
            printf("Error - pattern found more than once.\n");
            return NULL;
            }
         retval = base;
         }
      /* Increment base pointer */
      base++;
      }
   return retval;
   }


void do_help(void)
   {
   puts("\
kbs_patch - Patch kernel to support big SCSI drives\n\
\n\
Usage:\n\
    kbs_patch [1024 | 4096] <kernel_path>\n\
\n\
Examples:\n\
    kbs_patch /sau14/domain_os\n\
    kbs_patch 1024 /sau7/domain_os\n\
\n\
Description:\n\
    The Domain/OS kernel selected by <kernel_path> will be patched to support\n\
big SCSI drives.  This changes the code that calculates the number of blocks\n\
that the drive contains.  The pagesize argument is optional, and will default to\n\
the pagesize of the machine running the command.\n\
    The <kernel_path> argument must be the last one on the command line.  Only\n\
one file may be patched at a time.\n\
    Note that the patch that this program uses will not work on devices\n\
that have a block size that is not a multiple of 256 bytes.\n\
    See source code for technical details.\n\
\n\
Disclaimer:\n\
    No warranty for this program or its result are expressed or implied.  The\n\
author assumes no liability for data loss or any other damage caused by the use\n\
of this program.\n\
");
   }


int main(int argc, char *argv[])
   {
   int count;
   unsigned short *search_pattern, *replace_pattern;
   unsigned long search_length, replace_length;
   unsigned short *base, *patch_point;
   unsigned long pagesize;
   unsigned long length;
   struct stat stat_buf;
   status_$t status;

   /* say hello */
   printf("kbs_patch, version %s, built " __DATE__ " at " __TIME__ ".\n", Version);
   if (argc == 2)
      {
      pagesize = getpagesize();
      printf("Pagesize of machine is %d bytes.\n", pagesize);
      }
   else if (argc == 3)
      {
      pagesize = atol(argv[1]);
      printf("Selected page size is %d bytes.\n", pagesize);
      argv[1] = argv[2];
      }
   else
      {
      do_help();
      exit(EXIT_FAILURE);
      }
   /* Select the appropriate search/replace patterns, based on pagesize */
   switch (pagesize)
      {
      case 1024:
         search_pattern = search_pattern1k;
         search_length = search_pattern_length1k;
         replace_pattern = replace_pattern1k;
         replace_length = replace_pattern_length1k;
         break;
      case 4096:
         search_pattern = search_pattern4k;
         search_length = search_pattern_length4k;
         replace_pattern = replace_pattern4k;
         replace_length = replace_pattern_length4k;
         break;
      default:
         printf("Unrecognized page size of %d bytes, can't patch file.\n", pagesize);
         exit(EXIT_FAILURE);
      }
   printf(" Search pattern is");
   for (count = 0; count < search_length; count++)
      printf(" 0x%04x", search_pattern[count]);
   printf("\n Replace pattern is");
   for (count = 0; count < replace_length; count++)
      printf(" 0x%04x", replace_pattern[count]);
   puts("");

   /* Get the file length */
   if (stat(argv[1], &stat_buf))
      {
      perror("Can't get file size");
      exit(EXIT_FAILURE);
      }
   /* Map the file */
   printf("Mapping file '%s' for 0x%x bytes.\n", argv[1], stat_buf.st_size);
   base = ms_$mapl(argv[1], strlen(argv[1]), 0, (unsigned long)stat_buf.st_size, ms_$nr_xor_1w, ms_$wr, false, &length, &status);
   if (status.all != status_$ok)
      {
      printf("Error - Can't map file '%s'.\n", argv[1]);
      error_$print(status);
      exit(EXIT_FAILURE);
      }
   if (length == stat_buf.st_size)
      {
      /* Search for the patch point */
      patch_point = search(base, length/2, search_pattern, search_length);
      if (patch_point != NULL)
         {
         /* Update the file */
         for (count = 0; count < replace_length;count++)
            patch_point[count] = replace_pattern[count];
         printf("Pattern found, 0x%x Words changed at byte offset 0x%x.\n", replace_length, ((char *)patch_point)-((char *)base));
         }
      else
         {
         printf("Error - couldn't exactly one occurance of search pattern.\n");
         }
      }
   else
      {
      printf("Error - Wanted to map 0x%x bytes, only mapped 0x%x bytes.\n", stat_buf.st_size, length);
      exit(EXIT_FAILURE);
      }
   /* Unmap the file */
   ms_$unmap(base, length, &status);
   if (status.all != status_$ok)
      {
      printf("Error - Can't unmap file '%s'.\n", argv[1]);
      error_$print(status);
      exit(EXIT_FAILURE);
      }
   }

----------------------------------------------------------------------------------------

● NFS  2008/01 テスト

    $ wd /nix/internet/manual
    $ crd tmp
    
    かとうでロッグイン
    $ /etc/mount -o soft ic_sg2:/nix/cad /nix/internet/manual/tmp
    Must be root to use mount

    ル−トでロッグインし直してみた  Apollo の /etc/hosts に ic_sg2 がない。
    $ /etc/mount -o soft ic_sg2:/nix/cad /nix/internet/manual/tmp
    mount: ic_sg2 not in hosts database
    mount: giving up on:
       /nix/internet/manual/tmp

    Apollo の /etc/hosts に indy がある。
    $ /etc/mount -o soft indy:/nix/cad /nix/internet/manual/tmp

    $ ld tmp    INDY のファイルがみえている
    Directory "/sys/node_data/tmp":
    .X11-unix           lips_filter.config  nfsGlobalData
    nfs_data            print.u             rpcdep.dat
    rpcdllb.dat


    ic_sg2 2# cat /etc/exports
    # NFS exported filesystem database (see exports(4) for more information).
    # Entries in this file consist of lines containing the following fields:
    # filesystem    [ options ]     [ netgroup ] [ hostname ] ...
    #
    # Filesystem must be left-justified and may name any directory within a
    # local filesystem.  A backslash (\) at the end of a line permits splitting
    # long lines into shorter ones.  Netgroup(4) and hostname refer 
    # to machines or collections of machines to which filesystem is exported.
    #
    /nix -rw
    /usr1/indy/tenpu -rw


    $ /etc/mount -p
    /dev/wn768a      /                          5.3 rw           1187573952 0 0
    indy:/nix/cad /nix/internet/manual/tmp nfs soft,rw           1200269456 0 0

    解除
    $ /etc/umount /nix/internet/manual/tmp

    $ /etc/mount -p
    /dev/wn768a / 5.3 rw  1200269861 0 0


   * automount 起動

    $ /usr/etc/automount -m /indy /etc/auto.indy

    /etc/auto.indy
    -------------------------------------------
    |nix -rw,hard ic_sg2:/nix/cad
    |


   * INDY と Apollo のＮＦＳ

    ic_sg2 12# cat exports
    ----------------------------------
    |#
    |# NFS exported filesystem database (see exports(4) for more information).
    |#
    |# Entries in this file consist of lines containing the following fields:
    |#
    |# filesystem    [ options ]     [ netgroup ] [ hostname ] ...
    |#
    |# Filesystem must be left-justified and may name any directory within a
    |# local filesystem.  A backslash (\) at the end of a line permits splitting
    |# long lines into shorter ones.  Netgroup(4) and hostname refer 
    |# to machines or collections of machines to which filesystem is exported.
    |#
    |/nix/cad -rw

    # pwd
    //node_2cced/nix/internet/manual
    # mkdir abc

    # /etc/mount -o soft ic_sg2:/nix/cad /nix/internet/manual/abc

    $ wd abc;ld
    Directory "/nix/internet/manual/abc":

    .HSResource  bbb          com          doc          gyoumu       katou
    nix.ps       pp           sample       shibata      takagi       tmp
                                             
    解除

    $ /etc/umount /nix/internet/manual/abc

    $ /etc/mount -p     前
    /dev/wn0a / 5.3 rw  865570487 0 0

    $ /etc/mount -p     後
    /dev/wn0a          /                           5.3 rw                865570487 0 0
    ic_sg2:/nix/cad /nix/internet/manual/abc nfs soft,rw           865571367 0 0
            
    INDY 側の mount -p は変化なし

    $ /usr/etc/rpcinfo -p
       program vers proto   port
        100000    2   tcp    111  portmapper
        100000    2   udp    111  portmapper
        100024    1   udp    822  status
        100021    1   tcp    826  nlockmgr
        100024    1   tcp    824  status
        100021    1   udp   1025  nlockmgr
        100005    1   udp    829  mountd
        100005    2   udp    829  mountd
        100021    3   tcp    832  nlockmgr
        150001    1   udp    832  pcnfsd
        100005    1   tcp    833  mountd
        100021    3   udp   1026  nlockmgr
        100005    2   tcp    833  mountd
        100020    2   udp   1027  llockmgr
        100020    2   tcp    837  llockmgr
        100003    2   udp   2049  nfs
        100021    2   tcp    840  nlockmgr
        100021    2   udp   1029  nlockmgr

    $ /usr/ucb/netstat -a
    Active connections (including servers)
    Proto Recv-Q Send-Q  Local Address      Foreign Address    (state)
    tcp        0      0  localhost.960      localhost.sunrpc   TIME_WAIT
    tcp        0      0  node_2cced.1037    ic_sg2.telnet      ESTABLISHED


    ic_sg2 5# netstat -a 
    Active Internet connections (including servers)
    Proto Recv-Q Send-Q  Local Address          Foreign Address        (state)
    tcp        0      0  localhost.1276         localhost.sunrpc       TIME_WAIT   
    tcp        0      0  localhost.1275         localhost.sunrpc       TIME_WAIT   
    tcp        0      0  localhost.1274         localhost.sunrpc       TIME_WAIT   
    tcp        0      0  localhost.1273         localhost.sunrpc       TIME_WAIT   


    ダメな場合

    abc というディレクトリはない
    $ /etc/mount -o soft ic_sg2:/nix/cad /nix/internet/manual/abc
    mount: /nix/internet/manual/abc: No such file or directory

    $ wd /nix/internet/manual
    $ /etc/mount -o soft ic_sg2:/nix/cad abc
    mount: directory path must begin with '/'


● アポロとＳＵＮのＮＦＳの設定

   * ＳＵＮの設定

    /etc/export
    ------------------------------
    |/sun       ＳＵＮの /sun ディレクトリのみ解放する場合

    /etc/export
    ------------------------------
    |/          ＳＵＮの全ディレクトリを解放する場合
                                        
    % /usr/etc/expotfs -a   これをやると /etc/export が有効になる

   * アポロからＳＵＮへ

    $ /etc/mount -o soft sparc:/sun /sunp   開始
    $ /etc/mount -o soft sparc:/sun /sunpp
    $ /etc/mount -o soft sparc:/ /sun

    $ wd /;ld
    adus                 backup_log           bin
    bscom                bsd4.2               bsd4.3
    sunp                 sunpp     <<   仮想のディレクトリができている

    $ /etc/umount /sunpp                    終了
    $ /etc/umount /sunp


   * ＮＦＳを使う

    うまく行っている場合のアクル

    $ acl nfs?*
    Acl for nfsd:
    Required entries 
     root.%.%                         prwx-                 
     %.none.%                         [ignored]             
     %.%.none                         [ignored]             
     %.%.%                            -r-x-                 
     Extended entry rights mask:      -----

    Acl for nfsmount:
    Required entries 
     root.%.%                         prwx-                 
     %.none.%                         [ignored]             
     %.%.none                         [ignored]             
     %.%.%                            -r-x-                 
     Extended entry rights mask:      -----

    Acl for nfsstat:
    Required entries 
     root.%.%                         prwx-                 
     %.none.%                         [ignored]             
     %.%.none                         [ignored]             
     %.%.%                            -r-x-                 
     Extended entry rights mask:      -----

    Acl for nfsumount:
    Required entries 
     root.%.%                         prwx-                 
     %.none.%                         [ignored]             
     %.%.none                         [ignored]             
     %.%.%                            -r-x-                 
     Extended entry rights mask:      -----  


    /sys/node_data/etc/exports

    $ acl exports
    Acl for exports:
    Required entries 
     none.%.%                         [ignored]             
     %.none.%                         [ignored]             
     %.%.none                         [ignored]             
     %.%.%                            prwx-                 
     Extended entry rights mask:      -----


    ? Apollo 側に sun というディレクトリがあること

    ? $ /etc/mount -o soft sparc:/sun /sun

    ? $ wd /   ディレクトリ sun の中にいるとアンマウントできない
    ? $ /etc/umount /sun


● NFS のアカウントを作る

    ******************************************************************       
    ＮＦＳの設定のため person_id( user_id )とgroup_id を -2 にする方法
    ******************************************************************       

    $ wd /sys/node_data/tmp
    $ d
    Directory "/sys/node_data/tmp":
    .X11-unix     llbdbase.dat

    $ crd etc
    $ wd etc
    $ d
    Directory "/sys/node_data/tmp/etc":

    Empty directory.
    $ d

    Directory "/sys/node_data/tmp/etc":

    group  passwd
 
    group file
    --------------
    |nobody::-2:
    |testg::3000:
 
    passwd file
    --------------------------------------------
    |nobody::-2:-2:nobody for NFS dummy account:/
    |testp::3000:3000:test account for import passwd:/


    $ /etc/import_passwd -f -v -s /tmp/etc

    Preparing registry...

    Preparing foreign files...

    Creating Group entries from foreign group file...

    nobody::4294967294:
    >> Adding pgo entry for: nobody 4294967294

    testg::3000:
    >> Adding pgo entry for: testg 3000

    Creating Person entries and Accounts from foreign passwd file...

    nobody::4294967294:4294967294:nobody for NFS dummy account:/
    >> Adding pgo entry for: nobody 4294967294
    >> Adding account for nobody.nobody.none .
    
    testp::3000:3000:test account for import passwd:/
    >> Adding pgo entry for: testp 3000
    >> Adding account for testp.testg.none .

    Adding memberships from foreign group file...

    nobody::4294967294:

    testg::3000:

    0 name conflicts, 0 UNIX id conflicts, 0 unnamed groups, 0 errors, 0 warnings.

    Import operation successful.  Update registry [y/n]?  y

    Closing registry...

    Closing foreign files...

    Done.


    $ /etc/edrgy
    edrgy=> v nobody
    nobody[.nobody.none]::-2:-2:nobody for NFS dummy account:/::
    edrgy=> v testp
    testp[.testg.none]::3000:3000:test account for import passwd:/::
    edrgy=> do p
    Domain changed to: person
    edrgy=> v nobody -f
    nobody          -2                      5991304F.F0009BD8  %.%.%  pr --   
    edrgy=> v testp -f
    testp         3000                      59913050.10009BD8  %.%.%  pr --   
    edrgy=> do g
    Domain changed to: group
    edrgy=> v testg
    testg         3000
    edrgy=> v nobody -f
    nobody          -2                      5991304F.B0009BD8  %.%.%  pr -- l   ""  1992/06/25.15:22
    edrgy=> do a
    Domain changed to: account
    edrgy=> del testp.testg.none
    Please confirm delete of account "testp.testg.none" [y/n]: (n) y
    edrgy=> v testp
    ?(edrgy) Cannot retrieve any matching entries for "testp.%.%" - Entry not found (RGYC/Client)
    edrgy=> do p
    Domain changed to: person
    edrgy=> del testp
    Please confirm delete of primary name "testp" [y/n]: (n) y
    edrgy=> do g
    Domain changed to: group
    edrgy=> del testg
    ?(edrgy) Warning: any accounts for this group (%.testg.%) will also be deleted.
    Please confirm delete of primary name "testg" [y/n]: (n) y
    edrgy=> v testg
    ?(edrgy) "testg" - Entry not found (RGYC/Client)
    edrgy=> q