[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference aosg::lsm

Title:LSM
Moderator:SMURF::SHIDERLY
Created:Mon Jan 17 1994
Last Modified:Fri Jun 06 1997
Last Successful Update:Fri Jun 06 1997
Number of topics:803
Total number of notes:2852

769.0. "Can't see booted plex..." by NETRIX::"[email protected]" (John McDonald) Wed Mar 19 1997 18:08

I have a customer that has encaspulated (volencap) and mirrored
(volrootmirror)
his root and swap partitions. The commands completed successfully with no
errors. Now when he boots his system , he gets the message: 

LSM: Can't open device rz0a, device busy or inaccessible.

he's running Digital UNIX 4.0b (no patches)

The system boots fine, but the plex on the device that he boots from is
listed as 'DIABLED NODEVICE' in the volprint -hAt command, and a
voldisk list on the device shows 'errno: device path not valid'. The problem
is that if he boots from the other device (rz1), it's plex exhibits the same
symptoms but the system still boots fine.

Whichever device he boots from, the plex on that device in his rootvol
shows up as disabled, but the system runs fine.

The following is the volprint output when he booted from rz1:

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

Disk group: rootdg

DG NAME         GROUP-ID
DM NAME         DEVICE       TYPE     PRIVLEN  PUBLEN   PUBPATH
V  NAME         USETYPE      KSTATE   STATE    LENGTH   READPOL  PREFPLEX
PL NAME         VOLUME       KSTATE   STATE    LENGTH   LAYOUT   ST-WIDTH MODE
SD NAME         PLEX         PLOFFS   DISKOFFS LENGTH   DISK-NAME    DEVICE

dg rootdg       856714462.1025.falcon.netword.com

dm rz0a         rz0a         nopriv   0        251229   /dev/rrz0a
dm rz0b         rz0b         nopriv   0        484959   /dev/rrz0b
dm rz0d         rz0d         simple   1024     0        /dev/rrz0d
dm rz0g         rz0g         nopriv   0        1705059  /dev/rrz0g
dm rz1a         -            -        -        -        -
dm rz1b         rz1b         nopriv   0        484959   /dev/rrz1b
dm rz1d         rz1d         simple   1024     0        /dev/rrz1d
dm rz1g         rz1g         nopriv   0        1705059  /dev/rrz1g

v  rootvol      root         ENABLED  ACTIVE   251229   ROUND    -
pl rootvol-01   rootvol      DISABLED NODEVICE 251229   CONCAT   -        RW
sd rz1a-01p     rootvol-01   0        0        16       rz1a         -
sd rz1a-01      rootvol-01   16       16       251213   rz1a         -
pl rootvol-02   rootvol      ENABLED  ACTIVE   251229   CONCAT   -        RW
sd rz0a-01p     rootvol-02   0        0        16       rz0a         rz0a
sd rz0a-01      rootvol-02   16       16       251213   rz0a         rz0a

v  swapvol      swap         ENABLED  ACTIVE   484959   ROUND    -
pl swapvol-01   swapvol      ENABLED  ACTIVE   484959   CONCAT   -        RW
sd rz1b-01      swapvol-01   0        0        484959   rz1b         rz1b
pl swapvol-02   swapvol      ENABLED  ACTIVE   484959   CONCAT   -        RW
sd rz0b-01      swapvol-02   0        0        484959   rz0b         rz0b

v  vol-rz1g     fsgen        ENABLED  ACTIVE   1705059  SELECT   -
pl vol-rz1g-01  vol-rz1g     ENABLED  ACTIVE   1705059  CONCAT   -        RW
sd rz1g-01      vol-rz1g-01  0        0        1705059  rz1g         rz1g
pl vol-rz1g-02  vol-rz1g     ENABLED  ACTIVE   1705059  CONCAT   -        RW
sd rz0g-01      vol-rz1g-02  0        0        1705059  rz0g         rz0g

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

Notice that on the swap volume that was created, both plexes work fine.

The following is the voldisk list rz1a output from the same boot:

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

Device:    rz1a
devicetag: rz1a
type:      nopriv
info:      id=856714556.1086.falcon.netword.com
flags:     online error
pubpaths:  block=/dev/rz1a char=/dev/rrz1a
errno:     Device path not valid

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

I checked the device paths and they're all there. I also did a disklabel -r
on both devices. they're identical. here's what it looks like:

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

8 partitions:
#        size   offset    fstype   [fsize bsize   cpg]
  a:   251229        0     AdvFS                        # (Cyl.    0 - 182*)
  b:   484959   251229      swap                        # (Cyl.  182*- 535*)
  c:  4110480        0    unused     1024  8192         # (Cyl.    0 - 2987*)
  d:     1024   736188   LSMsimp                        # (Cyl.  535*- 535*)
  e:  1191936  1724416    unused     1024  8192         # (Cyl. 1253*- 2119*)
  f:  1194128  2916352    unused     1024  8192         # (Cyl. 2119*- 2987*)
  g:  1705059   737212     AdvFS                        # (Cyl.  535*- 1774*)
  h:  1668209  2442271     AdvFS                        # (Cyl. 1774*- 2987*)

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

The REALLY starnge part is that he did this on two different systems
in his ase, and they both exhibit the exact same behavior.

I'm going nuts trying to figure this one out. Any ideas?

John McDonald

[Posted by WWW Notes gateway]
T.RTitleUserPersonal
Name
DateLines
769.1some things to check ...NETRIX::"[email protected]"Thu Mar 20 1997 13:3221
It sounds like the rootvol is failing to start because
the underlying partition may have been already opened 
(e.g., the underlying partition was already directly 
mounted (opened) as root).

If this is the case, I would be interested in any additional 
data that might explain what went wrong or how they got into 
this position.   Anyways, here's some things to check that 
might help get them back on track:
   - Verify lsm_rootdev_is_volume is still set to 1 by
     typing "sysconfig -q lsm".
   - Verify the AdvFS volume within the AdvFS root domain 
     is a link to LSM's rootvol and not rz0a or rz1a by 
     typing "ls -l /etc/fdmns/root_domain".
   - Check /etc/fstab has the correct root domain specified.
   - If running in eager swap mode, check that /sbin/swapdefault
     is a link to the correct place.

Hope that helps.

[Posted by WWW Notes gateway]
769.2additioal info and need additional help..NETRIX::"[email protected]"decatl::johnsonThu Mar 20 1997 17:43166
Thanks for the quick response.   

Think your response has the key to the solution, however, 
I had found a couple items in which I have additional questions and
hopefully you may have clue as to the reason this occured.


following is 3 questions, issues::::

1.  sysconfigtab:

sysconfigtab is correct,  however, it is not being resolved correctly:

	falcon:/etc# grep -i lsm /etc/sysconfigtab
		lsm:
		lsm_rootdev_is_volume = 1
		lsm_swapdev_is_volume = 1
	falcon:/etc# 

	 
	falcon:/etc# sysconfig  -q lsm
		lsm:
		lsm_rootdev_is_volume = 0
		lsm_swapdev_is_volume = 0
		max-vol = 1024


I  dont have a lot of knowledge of the new (4.x) sysconfigtab structures, 
but appears that the sysconfigtab is corrupted (probably exiting in the
pci code), as the startup messages from uerf has the following. Dont
know how the corruption occured::

                                        Digital UNIX V4.0B  (Rev. 564); Tue 
                                         _Mar 18 10:37:26 EST 1997  
                                        physical memory = 192.00 megabytes. 
                                        available memory = 181.96 megabytes. 
                                        using 730 buffers containing 5.70 
                                         _megabytes of memory 
                                        Firmware revision: 4.7 
                                        PALcode: OSF version 1.21 
                                        AlphaServer 1000A 5/300 
                                        Mem_size = 199376896, consmem = 238 
                                        Sysconfigtab PCI_Option entry 28 
                                         _invalid. 
                                        Sysconfigtab PCI_Option entry 29 
                                         _invalid.                            
                                         
		...thru...                              ...thru...
				        Sysconfigtab PCI_Option entry 43 
                                         _invalid. 
                                        Sysconfigtab PCI_Option entry 44 
                                         _invalid. 


                                        pci0 at nexus 
                                        eisa0 at pci0 
                                        ace0 at eisa0 
                                        ace1 at eisa0 
                                        lp0 at eisa0 
                                        fdi0 at eisa0 
                                        fd0 at fdi0 unit 0 
                                        cirrus0 at eisa0 
                                        cirrus0: Cirrus Logic CL-GD5428 (SVGA)

                                         _512 Kbytes 
                                        Sysconfigtab PCI_Option entry 28 
                                         _invalid. 
                                        Sysconfigtab PCI_Option entry 30      
                                            
		...thru...                              ...thru... 

                                        Sysconfigtab PCI_Option entry 44 
                                         _invalid. 
                                        pci2000 at pci0 slot 8 aded 
                                         _by console) 
                                        scsi0 at isp0 slot 0 
                                        rz0 at scsi0 target 0 lun 0 (LID=0)  
                                         _(DEC     RZ28D    (C) DEC 0010) 
                                         _(Wide16) 
                                        rz1 at scsi0 target 1 lun 0 (LID=1) 
                                         _(DEC     RZ28D    (C) DEC 0010) 


/etc/sysconfigtab has several lines of PCI data, and appears to have
duplication
of eisa, pci, tc, etc entries.    The "Module_Config_Name", etc..
from the following errors are not defined in sysconfigtab, so am assuming
that the sysconfigtab customizations are exiting on pci and not getting to
LSM.

 
	falcon:/etc# sysconfig -q pci
		pci:
		Module_Config_Name = attribute does not allow this operation
		Module_Type = attribute does not allow this operation
		PCI_Option = attribute does not allow this operation
		PCI_Developer_Debug = 0
		Adapt_Config = 0
		Confl1 = 0
	falcon:/etc# 

RECOMMENDING:

	cp  /etc/sysconfigtab  /etc/sysconfigtab.corrupted
	cp  /etc/.new..sysconfigtab  /etc/sysconfigtab

	vi  /etc/sysconfigtab
		manually add the customizations, such as LSM...

and see if this resolves the issue.

will likely have to 
	#  voldg  adddisk rz1a=rz1a

Additional questions:

2.   fstab has the following entry, is the '1  1' ignored?  _______________

		root_domain#root /      advfs rw 1 1


    falcon:/# ls -l /etc/fdmns/root_domain

	lrwxrwxrwx   1 root     system        23 Feb 23 11:14 rootvol -> 
		/dev/vol/rootdg/rootvol
 
    falcon:/# ls -la /sbin/swapdefault
	lrwxrwxrwx   1 root     system        23 Feb 23 11:14 /sbin/swapdefault
		 -> /dev/vol/rootdg/swapvol



2.   The error
	'errno: device path not valid' is from the lsm configuration
	database, 

	will this resolve itself using the voldg command above?  ______


     The following shows the invalid path (from volprint)::

    dm   rz1a
        tutil0="
        tutil1="
        tutil2="
        da_name=
        pub_bdev=0/0
        priv_bdev=0/0


	
    dm   rz0a
        tutil0="
        tutil1="
        tutil2="
        da_name=rz0a
        device_tag=rz0
        da_type=nopriv
        pub_bpath="/dev/rz0a
        pub_cpath="/dev/rrz0a


Sid Johnson
Customer Support/Atlanta
[Posted by WWW Notes gateway]