ITEM: H2112L

Filesystem sizes of df and smit do not match.


Question:

I have a RS6000 running 3.2.5, and I am having some filesystem
problems.  I have created three journaled file systems on a new hdisk.
When I look at the size of the filesystem through a df command, it
shows a size of 303104 Kbytes, and about 39% used.  When I try to look
at the size of the filesystem through smit, it shows a size of zero.
When I do a lsfs /\, I get:

Name            Nodename   Mount Pt            VFS   Size Options 
/dev/db           --           ?               jfs   ----   --

What could be the problem?

Response:

I had him run some commands to get the status of his filesystems. "df"
seems to return the correct information. He has 3 hard disks that are
all in rootvg.  I had him run "lsvg -l rootvg" to get a list of all
the filesystems and Logical Volumes (LVs).  He had all the filesystems 
showing, but the LV names were not ones he had for the filesystems. 
He had a filesystem called /db with a LV of lv05. "lsvg" showed it had 
an lvname of db. He had the same situation for a filesystem called 
/usr/asql. The LV got changed to asql in the "lsvg" list. I then went 
to check the lower level stuff. I had him run "lsdev -Cc disk" to get
a list of the disks" He had four showing, but one was only "defined".

        hdisk0  Available       400meg
        hdisk1  Available       400meg
        hdisk2  Defined         400meg
        hdisk3  Available       Other Scsi Disk

hdisk1 and hdisk2 showed the same address. I then had him run "lspv -l
hdisk2" and it showed LVs in it. To be safe, I had him run "lqeuryvg -p
\ -Pt" on all the available disks. The command showed all the
right number of entries (3) but it failed on hdisk3. At this point, I
felt it would be best to try to work from the bottom up. I sent him
the fax for DISKSYNC. 

Customer had restored the configuration for the old machine from tape 
after setting everything up (/etc/objrepos for one). This could 
definitely cause these types of problems. I told him that it would be 
likely that they would need to run RVGRECOVER as well. 

Response:

Customer ran the DISKSYNC script and was successful. They now only had 
3 disks showing and all were available. I had him run "lqueryvg -p
hdiskX -AT" on all the hdisks and they showed the correct information. 
He still had the same filesystem problems, but the RVGRECOVER script
should handle this.  You ran the RVGRECOVER script and it fixed the
system.

In the future, please do not restore EVERYTHING from your system 
backups.  Only restore the directories you need. In this case, you were
restoring from "/".

Response:

Here is the DISKSYNC script:
                                                                 04/21/93
  SYNCHRONIZING DISK NAMES
        NOTE:  The problem for which you received this document is
       not considered a code warranty issue.  This document is pro-
       vided as an aid by the Austin AIX Support Center.  If you
       need further assistance, contact your local branch office or
       point of sale, or call 1-800-CALL-AIX for information about
       support offerings.  All of the above services may be
       billable.
        Comments about this document may be sent by fax to "Info
       Feedback" at (512) 908-3810.  IBM representatives can send
       comments internally to ROUSHC at AUSVM8.

       DISCLAIMER:  The information contained in this document is 
       distributed "AS IS" without any warranties of any kind either 
       express or implied.  IBM will not be responsible for any direct, 
       incidental, consequential, special or indirect damages.  IBM 
       EXPRESSLY DISCLAIMS ANY IMPLIED WARRANTY OF MERCHANTABILITY AND 
       ANY IMPLIED WARRANTY OF FITNESS FOR A PARTICULAR PURPOSE.

       The use of this information or the implementation of any of 
       these techniques is the sole responsibility of the customer and 
       depends on the customer's ability to evaluate and integrate this 
       information or implementation into the customer's operational 
       environment.
       ----------------------------------------------------------------
        Use the following script when the names of your hard disks are 
       out of order (for example:  hdisk0, hdisk2, hdisk3 instead of 
       hdisk0, hdisk1, hdisk2).  The order of the disk names generally 
       does not cause errors, but it may cause confusion for the user.  
       Run the following "dsksync" script to alleviate such confusion.  
       The script will rename the hard disks.
        This document applies to AIX 3.1 and 3.2 on the RISC System/6000.
        Make sure the key is in "Normal" position before running this 
       script.
   ----------------------------------------------------------------------
        lsdev -Cc disk | awk '{ print $1 }' | while read HDname; do
                 odmdelete -q "name = $HDname" -o CuAt
                 odmdelete -q "value = $HDname" -o CuAt
                 odmdelete -q "name = $HDname" -o CuDep
                 odmdelete -q "name = $HDname" -o CuDv
                 odmdelete -q "value3 = $HDname" -o CuDvDr
                 odmdelete -q "name = $HDname" -o CuVPD
       done
       rm -f /dev/hdisk*
       rm -f /dev/rhdisk*
        savebase
   ----------------------------------------------------------------------
 |           When the shell script completes successfully, run the
|           following command to shutdown and reboot.
 |                shutdown -Fr
                                              Synchronizing Disk Names  1

--> End of disksync script
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Begin RVGRECOVER

 
                                                                07/28/93

  RECOVERING VOLUME GROUPS IN AIX 3.1 AND 3.2
        NOTE:  The problem for which you received this document is
       not considered a code warranty issue.  This document is pro-
       vided as an aid by the Austin AIX Support Center.  If you
       need further assistance, contact your local branch office or
       point of sale, or call 1-800-CALL-AIX for information about
       support offerings.  All of the above services may be
       billable.  Faxes on a variety of subjects may be ordered 
       free of charge from 1-800-IBM-4FAX.
        Comments about this document may be sent by fax to "Info
       Feedback" at (512) 823-7634.  IBM representatives can send
       comments internally to ROUSHC at AUSVM8.

       DISCLAIMER:  The information contained in this document is 
       distributed "AS IS" without any warranties of any kind either 
       expressed or implied.  IBM will not be responsible for any direct, 
       incidental, consequential, special or indirect damages.  IBM 
       EXPRESSLY DISCLAIMS ANY IMPLIED WARRANTY OF MERCHANTABILITY AND 
       ANY IMPLIED WARRANTY OF FITNESS FOR A PARTICULAR PURPOSE.

       The use of this information or the implementation of any of 
       these techniques is the sole responsibility of the customer and 
       depends on the customer's ability to evaluate and integrate this 
       information or implementation into the customer's operational 
       environment.
   About this Document

       Run the script in this document when the ODM (Object Data 
       Manager) entries for the root volume group are corrupted.  The 
       script may be modified for volume groups other than rootvg.  
       (See "About the Script" in this document.)
   Problem Determination

       When the ODM entries for a volume group are corrupted, you may 
       notice that volume-group and logical-volume commands fail, that 
       you cannot change the size of file systems, or that the system 
       cannot find volume groups or device IDs.  Run "lslv -l" to see 
       if a corrupted ODM description is the problem -- if it is, 
       information will probably be missing from the "lslv" output, or
       there may be "?"s in the output.

  Before Using the Script

       Before using the following script, save the current version of 
       your /etc/objrepos/Cu* files, in case you want to go back to 
       that version at some time.  (In the following commands, replace 
       "mmmdd" with the month and day, such as "jan01".)

          cd /etc/objrepos
          cp CuAt cuat.mmmdd      
          cp CuDep cudep.mmmdd
          cp CuDv cudv.mmmdd
          cp CuDvDr cudvdr.mmmdd

  About the Script
 |      1.  This script contains only a few provisions to handle error 
           conditions because it is designed to minimize typing mistakes.
        2.  The PV (physical volume) and VG (volume group) variables on 
|          the first two lines of the script are defaulted for the root 
|          volume group (rootvg).  The PV is set to /dev/ipldevice, which
|          is a synonym for one of the physical volumes that is a member
|          of the rootvg volume group.

           If you wish to run this script for a volume group other than 
|          the rootvg, then you must change the VG variable to match the 
|          volume group you wish to fix, and change the PV variable to 
|          match any one of the physical volumes (/dev/hdisk\#) that is a 
|          member of that volume group.  To determine to which VG a PV
|          belongs, enter the command:

|             lsvg `lqueryvg -p hdisk\# -v` | grep GROUP
        3.  It is not necessary to reboot after running this script.  
   The Script

       Run the following script while in Normal mode and logged in as root.  
       On line 14 of the script, "[" should be a left bracket and "]" 
       should be a right bracket.
   ----------------------------------------------------------------------
        PV=/dev/ipldevice
       VG=rootvg
        lqueryvg -p $PV -L | awk '{ print $2 }' | while read LVname; do
                odmdelete -q "name = $LVname" -o CuAt
                odmdelete -q "name = $LVname" -o CuDv
                odmdelete -q "value3 = $LVname" -o CuDvDr
       done
       odmdelete -q "name = $VG" -o CuAt
       odmdelete -q "parent = $VG" -o CuDv
       odmdelete -q "name = $VG" -o CuDv
       odmdelete -q "name = $VG" -o CuDep
       odmdelete -q "dependency = $VG" -o CuDep
       if [ "$VG" = rootvg ]
       then
         odmdelete -q "value1 = 10" -o CuDvDr
       else
         odmdelete -q "value1 = $VG" -o CuDvDr
       fi
       odmdelete -q "value3 = $VG" -o CuDvDr

       importvg -y $VG $PV             \# ignore lvaryoffvg errors
       varyonvg $VG        
   ----------------------------------------------------------------------
   END OF DOCUMENT (rvgrecover.krn)


Support Line: Filesystem sizes of df and smit do not match. ITEM: H2112L
Dated: April 1994 Category: N/A
This HTML file was generated 99/06/24~13:30:48
Comments or suggestions? Contact us