Re: set the number of inodes during FS creation via pressed

From	phoebus phoebus <frphoebus@yahoo.fr>
Newsgroups	linux.debian.user
Subject	Re: set the number of inodes during FS creation via pressed
Date	2021-11-14 15:00 +0100
Message-ID	<Djod3-7Wr-9@gated-at.bofh.it> (permalink)
References	<DiPfk-3FU-7@gated-at.bofh.it> <DiPfk-3FU-5@gated-at.bofh.it> <Dj4R4-4P7-5@gated-at.bofh.it> <Dja0p-7MI-1@gated-at.bofh.it>
Organization	linux.* mail to news gateway

Show all headers | View raw

[Multipart message — attachments visible in raw view] - view raw

 Thanks for the update.
 >>From there I found this file for bullseye:>>  https://preseed.debian.net/debian-preseed/bullseye/amd64-main-full.txt>>which contains this stanza which appears to match your description:
>>### Description: Typical usage of this partition:
>>#  Please specify how the file system is going to be used, so that>>#  optimal file system parameters can be chosen for that use.>>#  .>>#  standard = standard parameters,>>#  news = one inode per 4KB block,>>#  largefile = one inode per megabyte,>>#  largefile4 = one inode per 4 megabytes.>># d-i partman-basicfilesystems/specify_usage select <choice>>># Possible choices: ${CHOICES}>>>>Which is not what the OP requested, but maybe it helps.  It is the only>>occurrence of "inode" in that file.  I did not check the other two files>>that contain the "contrib" and "non-free" data.
I tryed to adapt like this with this option: **specify_usage{  }**
--------------------------------------------------------------------------------------------------------------

    1024 1024 2048 ext4 \      lv_name{ var+log } \
      method{ lvm } format{ } \
        use_filesystem{ } filesystem{ ext4 } \
      mountpoint{ /var/log } \
        options/nodev{ nodev } \
        options/nosuid{ nosuid } \
        options/noexec{ noexec } \
      $lvmok{ } \
    . \
    1024 1024 2048 ext4 \
      lv_name{ var+loga } \
      method{ lvm } format{ } \
        use_filesystem{ } filesystem{ ext4 } \
      mountpoint{ /var/loga } \
        options/nodev{ nodev } \
        options/nosuid{ nosuid } \
        options/noexec{ noexec } \
        specify_usage{ news } \
      $lvmok{ } \
    . \
    1024 1024 2048 ext4 \
      lv_name{ var+logb } \
      method{ lvm } format{ } \
        use_filesystem{ } filesystem{ ext4 } \
      mountpoint{ /var/logb } \
        options/nodev{ nodev } \
        options/nosuid{ nosuid } \
        options/noexec{ noexec } \
        specify_usage{ large } \
      $lvmok{ } \
    . \
    1024 1024 2048 ext4 \
      lv_name{ var+logc } \
      method{ lvm } format{ } \
        use_filesystem{ } filesystem{ ext4 } \
      mountpoint{ /var/logc } \
        options/nodev{ nodev } \
        options/nosuid{ nosuid } \
        options/noexec{ noexec } \
        specify_usage{ large4 } \
      $lvmok{ } \
    . \
    1024 1024 2048 ext4 \
      lv_name{ var+logd } \
      method{ lvm } format{ } \
        use_filesystem{  } filesystem{ ext4 } \
      mountpoint{ /var/logd } \
        options/nodev{ nodev } \
        options/nosuid{ nosuid } \
        options/noexec{ noexec } \
        specify_usage{ standard } \
      $lvmok{ } \
    . \
--------------------------------------------------------------------------------------------------------------

But this change nothing after verfication the bytes per inode is 16k in all case. (<bytes-per-inode> = <block count> x <block size> / <inode count>)--------------------------------------------------------------------------------------------------------------
 root@debian:~# dumpe2fs /dev/mapper/rootvg-var+log | grep -E "(Block count|Block size|Inode count)"
dumpe2fs 1.46.2 (28-Feb-2021)
Inode count:              62464
Block count:              249856
Block size:               4096
root@debian:~# dumpe2fs /dev/mapper/rootvg-var+loga | grep -E "(Block count|Block size|Inode count)"
dumpe2fs 1.46.2 (28-Feb-2021)
Inode count:              62464
Block count:              249856
Block size:               4096
root@debian:~# dumpe2fs /dev/mapper/rootvg-var+logb | grep -E "(Block count|Block size|Inode count)"
dumpe2fs 1.46.2 (28-Feb-2021)
Inode count:              62464
Block count:              249856
Block size:               4096
root@debian:~# dumpe2fs /dev/mapper/rootvg-var+logc | grep -E "(Block count|Block size|Inode count)"
dumpe2fs 1.46.2 (28-Feb-2021)
Inode count:              62464
Block count:              249856
Block size:               4096
root@debian:~# dumpe2fs /dev/mapper/rootvg-var+logd | grep -E "(Block count|Block size|Inode count)"
dumpe2fs 1.46.2 (28-Feb-2021)
Inode count:              62464
Block count:              249856
Block size:               4096
root@debian:~#--------------------------------------------------------------------------------------------------------------

Tryed also with 
-----------------------------------------------------------------------------------------------------------d-i partman-basicfilesystems/specify_usage select news
----------------------------------------------------------------------------------------------------------

And ugly workaround will be to update inode_ration into  /etc/mke2fs.conf with sed  via 

-------------------------------------------------------------------------------------------------------------------------
d-i partman/early_command string \    sed -i 's/inode_ratio = 16384/inode_ratio = 4096/g' /etc/mke2fs.conf
------------------------------------------------------------------------------------------------------------------------

but it apply to all FS.

/etc/mke2fs.con also indicates the values:
--------------------....
    news = {
        inode_ratio = 4096
    }
    largefile = {
        inode_ratio = 1048576
        blocksize = -1
    }
    largefile4 = {
        inode_ratio = 4194304
        blocksize = -1
    }
.....

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

I'll switch from EXT4 to XFS to not be worried about inodes defined at creation.
Unfortunaltely XFS don't take in charge the reduction but only the increase of FS size.
Don't be sure possibe to do "thin provisionning" and "thin pool" => Configure Thin Provision LVM with pressed as can be done with kickstarrt into RHEL.

Regards,
Thierry

Thread

set the number of inodes during FS creation via pressed phoebus phoebus <frphoebus@yahoo.fr> - 2021-11-13 01:40 +0100
  Re: set the number of inodes during FS creation via pressed Stanislav Vlasov <stanislav.v.v@gmail.com> - 2021-11-13 11:20 +0100
  Re: set the number of inodes during FS creation via pressed David Wright <deblis@lionunicorn.co.uk> - 2021-11-13 18:20 +0100
    Re: set the number of inodes during FS creation via pressed David <bouncingcats@gmail.com> - 2021-11-13 23:50 +0100
      Re: set the number of inodes during FS creation via pressed phoebus phoebus <frphoebus@yahoo.fr> - 2021-11-14 15:00 +0100
        Re: set the number of inodes during FS creation via pressed David Wright <deblis@lionunicorn.co.uk> - 2021-11-14 18:00 +0100
        Re: set the number of inodes during FS creation via pressed David <bouncingcats@gmail.com> - 2021-11-15 01:50 +0100
          Re: set the number of inodes during FS creation via pressed phoebus phoebus <frphoebus@yahoo.fr> - 2021-11-15 12:00 +0100

csiph-web