LinuxQuestions.org - LVM: Building a LV and a Filesystem to just fit

- Linux - General (https://www.linuxquestions.org/questions/linux-general-1/)

- - LVM: Building a LV and a Filesystem to just fit (https://www.linuxquestions.org/questions/linux-general-1/lvm-building-a-lv-and-a-filesystem-to-just-fit-599096/)

LVM: Building a LV and a Filesystem to just fit

The idea: Write a script that will analyze a directory and build a LV/FS to fit that directory perfectly with no wasted space.

The problem: Getting the sizing right. Default settings make too many inodes and waste space. Analyzing the directory with find just seems to get too close and then some overhead of creating the FS makes it just not fit either on inodes or space. In general it seems more art than algorithm.

I'm using RHEL4 and LVM2.

:-Dan

the style and content of this reads just like a university final year project to me. as such we aren't here to do your wrk for you, especially as you've done nothing but recite it, and haven't even asked a question at all...

Thanks I guess. I never had much university and certainly never made it to any final year. At least we agree that the problem is challenging. I had considered posting all my trial and error methods, but it is really just guess work and I was concerned that my attempts would tempt someone to make minor tweaks to my trail and failures, instead of making an unprejudiced suggestion.

I'm not looking for anyone to do my work for me, I would like someone who understands the interactions between the LVM and the filesystem to point me in the right direction.

Your best bet is to post up what you have done already. Otherwise we'll end up suggesting things you have tried. It also may be that you were on the right track with one of your ideas. We'll never know unless you tell us.

Script and Eaxmples

This is my trial and error script and a few examples. I'm sure the trick is in figuring out some factor to add to the size and inode counts, but I'd like it to be based on reason and not trial and error.

Code:

    1  #!/bin/bash

    2

    3  DIR=${1}

    4  NAME=${DIR##*/}

    5

    6  if [[ ! -d ${DIR} ]]; then

    7          echo "[${DIR}] is not a directory"

    8          exit 2

    9  fi

    10

    11  declare -i total=0 cnt=0

    12  while read blocks;do

    13          (( ++cnt ))

    14          (( total += blocks ))

    15  done < <( /usr/bin/find ${DIR} -printf "%k\n" )

    16

    17  echo "inodes=${cnt}, 1kBlocks=${total}"

    18

    19  set -x

    20  lvcreate --name ${NAME} --size ${total}k ReadOnly

    21  [[ $? -ne 0 ]] && exit 1

    22  mke2fs -N ${cnt} -L ${NAME} /dev/ReadOnly/${NAME}

    23  [[ $? -ne 0 ]] && exit 1

    24  mkdir /testing/${NAME}

    25  mount /dev/ReadOnly/${NAME} /testing/${NAME}

    26  [[ $? -ne 0 ]] && exit 1

    27  df -P /testing/${NAME};df -Pi /testing/${NAME}

    28  cd ${DIR}

    29  tar cf - . | (cd /testing/${NAME};tar xf -)

    30  df -P /testing/${NAME};df -Pi /testing/${NAME}

    31  echo "umount /testing/${NAME};lvremove /dev/ReadOnly/${NAME}"

This run is small (and it fails). The actual volumes we expect to create are in the 60-80GB range (but the runs take a long time and always fail with this script:

Code:

# ./MakeImgVol.sh  /cacheimg/raidu0/pcl/v007

inodes=205, 1kBlocks=9716

+ lvcreate --name v007 --size 9716k ReadOnly

  Rounding up size to full physical extent 12.00 MB

  Logical volume "v007" created

+ [[ 0 -ne 0 ]]

+ mke2fs -N 205 -L v007 /dev/ReadOnly/v007

mke2fs 1.35 (28-Feb-2004)

Filesystem label=v007

OS type: Linux

Block size=1024 (log=0)

Fragment size=1024 (log=0)

208 inodes, 12288 blocks

614 blocks (5.00%) reserved for the super user

First data block=1

Maximum filesystem blocks=12582912

2 block groups

8192 blocks per group, 8192 fragments per group

104 inodes per group

Superblock backups stored on blocks:

        8193



Writing inode tables: done

Writing superblocks and filesystem accounting information: done



This filesystem will be automatically checked every 36 mounts or

180 days, whichever comes first.  Use tune2fs -c or -i to override.

+ [[ 0 -ne 0 ]]

+ mkdir /testing/v007

mkdir: cannot create directory `/testing/v007': File exists

+ mount /dev/ReadOnly/v007 /testing/v007

+ [[ 0 -ne 0 ]]

+ df -P /testing/v007

Filesystem        1024-blocks      Used Available Capacity Mounted on

/dev/mapper/ReadOnly-v007    12253      108    11531      1% /testing/v007

+ df -Pi /testing/v007

Filesystem            Inodes  IUsed  IFree IUse% Mounted on

/dev/mapper/ReadOnly-v007    208      11    197    6% /testing/v007

+ cd /cacheimg/raidu0/pcl/v007

+ tar cf - .

+ cd /testing/v007

+ tar xf -

tar: ./v00707700v/img2c914.tif: Cannot open: No space left on device

tar: ./v00705400v: Cannot mkdir: No space left on device

tar: ./v00705400v/img2de5e.tif: Cannot open: No such file or directory

tar: ./v00705400v/img2ded2.tif: Cannot open: No such file or directory

tar: ./v00705400v/img2df3a.tif: Cannot open: No such file or directory

tar: ./v00705400v/img2dfe9.tif: Cannot open: No such file or directory

tar: ./v00705400v/img2e027.tif: Cannot open: No such file or directory

tar: Error exit delayed from previous errors

+ df -P /testing/v007

Filesystem        1024-blocks      Used Available Capacity Mounted on

/dev/mapper/ReadOnly-v007    12253      9134      2505      79% /testing/v007

+ df -Pi /testing/v007

Filesystem            Inodes  IUsed  IFree IUse% Mounted on

/dev/mapper/ReadOnly-v007    208    208      0  100% /testing/v007

+ echo 'umount /testing/v007;lvremove /dev/ReadOnly/v007'

umount /testing/v007;lvremove /dev/ReadOnly/v007

Success with a larger volume:

Code:



# ./MakeImgVol.sh  /cacheimg/raidu0/pcl/v000

inodes=6417, 1kBlocks=399540

+ lvcreate --name v000 --size 399540k ReadOnly

  Rounding up size to full physical extent 392.00 MB

  Logical volume "v000" created

+ [[ 0 -ne 0 ]]

+ mke2fs -N 6417 -L v000 /dev/ReadOnly/v000

mke2fs 1.35 (28-Feb-2004)

Filesystem label=v000

OS type: Linux

Block size=1024 (log=0)

Fragment size=1024 (log=0)

6664 inodes, 401408 blocks

20070 blocks (5.00%) reserved for the super user

First data block=1

Maximum filesystem blocks=67633152

49 block groups

8192 blocks per group, 8192 fragments per group

136 inodes per group

Superblock backups stored on blocks:

        8193, 24577, 40961, 57345, 73729, 204801, 221185



Writing inode tables: done

Writing superblocks and filesystem accounting information: done



This filesystem will be automatically checked every 35 mounts or

180 days, whichever comes first.  Use tune2fs -c or -i to override.

+ [[ 0 -ne 0 ]]

+ mkdir /testing/v000

+ mount /dev/ReadOnly/v000 /testing/v000

+ [[ 0 -ne 0 ]]

+ df -P /testing/v000

Filesystem        1024-blocks      Used Available Capacity Mounted on

/dev/mapper/ReadOnly-v000    400452      2062    378320      1% /testing/v000

+ df -Pi /testing/v000

Filesystem            Inodes  IUsed  IFree IUse% Mounted on

/dev/mapper/ReadOnly-v000    6664      11    6653    1% /testing/v000

+ cd /cacheimg/raidu0/pcl/v000

+ tar cf - .

+ cd /testing/v000

+ tar xf -

+ df -P /testing/v000

Filesystem        1024-blocks      Used Available Capacity Mounted on

/dev/mapper/ReadOnly-v000    400452    384437        0    100% /testing/v000

+ df -Pi /testing/v000

Filesystem            Inodes  IUsed  IFree IUse% Mounted on

/dev/mapper/ReadOnly-v000    6664    6427    237  97% /testing/v000

+ echo 'umount /testing/v000;lvremove /dev/ReadOnly/v000'

umount /testing/v000;lvremove /dev/ReadOnly/v000

Any file system gurus out there?

Here is an example of failure with a larger volume and a small tweak to the mke2fs command (on line 22) in the script to reduce the blocks "reserved for the super user" from the default 5% to 0% (didn't seem to make any difference):

22 mke2fs -m 0 -O sparse_super,filetype -N ${cnt} -L ${NAME} /dev/ReadOnly/${NAME}

The man page says that "sparse_super,filetype" are defaults. Just making sure.

Note: The size of this volume is in the range we expect to see.

Code:

# ./MakeImgVol.sh  /cacheimg/raidu0 

inodes=1122588, 1kBlocks=69711504

+ lvcreate --name raidu0 --size 69711504k ReadOnly

  Rounding up size to full physical extent 66.48 GB

  Logical volume "raidu0" created

+ [[ 0 -ne 0 ]]

+ mke2fs -m 0 -O sparse_super,filetype -N 1122588 -L raidu0 /dev/ReadOnly/raidu0

mke2fs 1.35 (28-Feb-2004)

Filesystem label=raidu0

OS type: Linux

Block size=4096 (log=2)

Fragment size=4096 (log=2)

1123584 inodes, 17428480 blocks

0 blocks (0.00%) reserved for the super user

First data block=0

Maximum filesystem blocks=20971520

532 block groups

32768 blocks per group, 32768 fragments per group

2112 inodes per group

Superblock backups stored on blocks:

        32768, 98304, 163840, 229376, 294912, 819200, 884736, 1605632, 2654208,

        4096000, 7962624, 11239424



Writing inode tables:  done

Writing superblocks and filesystem accounting information: done



This filesystem will be automatically checked every 37 mounts or

180 days, whichever comes first.  Use tune2fs -c or -i to override.

+ [[ 0 -ne 0 ]]

+ mkdir /testing/raidu0

mkdir: cannot create directory `/testing/raidu0': File exists

+ mount /dev/ReadOnly/raidu0 /testing/raidu0

+ [[ 0 -ne 0 ]]

+ df -P /testing/raidu0

Filesystem        1024-blocks      Used Available Capacity Mounted on

/dev/mapper/ReadOnly-raidu0  69568904    53272  69515632      1% /testing/raidu0

+ df -Pi /testing/raidu0

Filesystem            Inodes  IUsed  IFree IUse% Mounted on

/dev/mapper/ReadOnly-raidu0 1123584      11 1123573    1% /testing/raidu0

+ cd /cacheimg/raidu0

+ tar cf - .

+ cd /testing/raidu0

+ tar xf -

tar: ./pcl/v102/v10271900v/imge126a.tif: Wrote only 4608 of 10240 bytes

tar: Skipping to next header

tar: ./pcl/v102/v10271900v/imge1423.tif: Cannot write: No space left on device

tar: Skipping to next header

tar: ./pcl/v102/v10271900v/imge156d.tif: Cannot write: No space left on device

tar: Skipping to next header

... over 4,000 messages like this ...

tar: ./pcl/v988/v98816750/imgd1ca9.tif: Cannot open: No such file or directory

tar: ./pcl/v988/v98816750/imgd1d94.tif: Cannot open: No such file or directory

tar: Error exit delayed from previous errors

+ df -P /testing/raidu0

Filesystem        1024-blocks      Used Available Capacity Mounted on

/dev/mapper/ReadOnly-raidu0  69568904  69568904        0    100% /testing/raidu0

+ df -Pi /testing/raidu0

Filesystem            Inodes  IUsed  IFree IUse% Mounted on

/dev/mapper/ReadOnly-raidu0 1123584 1118460    5124  100% /testing/raidu0

+ echo 'umount /testing/raidu0;lvremove /dev/ReadOnly/raidu0'

umount /testing/raidu0;lvremove /dev/ReadOnly/raidu0