[LU-11060] mkfs.lustre cli command not supported in 2.11

• Status: Resolved
• Project: Lustre
• Affects Version/s: Lustre 2.11.0
• Type: Bug
• Priority: Major
• Reporter: Abe
• Assignee: WC Triage
• Resolution: Done
• Severity: 3

Description

the mkfs.lustre cli is not supported under 2.11 release of lustre:

Below is the list of components installed on the server:

 dkms status

lustre-all, 2.11.0, 3.10.0-693.21.1.el7_lustre.x86_64, x86_64: installed

spl, 0.7.6, 3.10.0-693.21.1.el7_lustre.x86_64, x86_64: installed

zfs, 0.7.6, 3.10.0-693.21.1.el7_lustre.x86_64, x86_64: installed

 

we are able to create the zfs pools:

 zpool create -f -O canmount=off -o cachefile=none mgspool mirror sdb sdc

zpool status

  pool: mgspool

 state: ONLINE

  scan: none requested

config:

 

NAME        STATE     READ WRITE CKSUM

mgspool     ONLINE       0     0     0

  mirror-0  ONLINE       0     0     0

    sdb     ONLINE       0     0     0

    sdc     ONLINE       0     0     0

 however the mfs.lustre cli fails:

 

mkfs.lustre --mgs --fsname=$NAME --reformat --servicenode=$MGS_NID --servicenode=$MDT_NID --mgsnode=$MGS_NID --mgsnode=$MDT_NID --backfstype=zfs mgspool/mgt

mkfs.lustre: command not found!!

 

Has this been deprecated or is there something missing ?

 

thanks,

Abe

 

 

 

 

Comments

[Andreas Dilger]

The mkfs.lustre tool is still in use and needs to be installed from the RPL. Almost certainly something is missing. This is a standard command that is still in use.

You can try "which mkfs.lustre" to check if it is in your path, or "rpm -ql <package>"

[Abe]

You are right Andreas,

 

the lustre rpm was not installed although dams shows as installed:

 

yum install lustre-2.11.0-1.el7.x86_64.rpm    

mkfs.lustre

 modprobe lustre

 modprobe zfs

 modprobe lnet

  lctl network up

 

and it is working now:

mkfs.lustre --mgs --fsname=$NAME --reformat --servicenode=$MGS_NID --servicenode=$MDT_NID --mgsnode=$MGS_NID --mgsnode=$MDT_NID --backfstype=zfs mgspool/mgt

 

mgspool/mgt on /mnt/lustre/mgt type lustre (ro,seclabel,svname=MGS,nosvc,mgs,osd=osd-zfs)

 

Also do you know of documentation on how to setup data on metadata feature in 2.11 release ?

 

Thanks & Best Regards,

Abe

 

 

 

[Andreas Dilger]

See https://jira.hpdd.intel.com/secure/attachment/28729/DOM_user_guide.pdf

[Abe]

thanks Andreas for providing info this, one more question are there any default settings already for the strip size etc or does the user has to set these params before using metadata servers ?

 

Best regards,

Abe

[Andreas Dilger]

There are several levels of default values for file layouts. In order of increasing priority

• at the lowest level are the default values that are compiled into the code if no other values are available, namely stripe_count=1 and stripe_size=1MB (which only really matters for multi-stripe files)
• there are default values that can be specified at format time with -param=lov.stripesize=1M and -param=lov.stripecount=1 (or whatever) which apply to all files unless some other layout is specified
• there may be a default layout stored on the filesystem root directory using lfs setstripe ... <mountpoint> (typically by the sysadmin) to allow plain (RAID-0) or composite (PFL) layouts to be specified for all new files created in the filesystem that don't have any other layout
• there may be a default layout stored on the parent directory (normally by the user) using lfs setstripe ... <parent_directory> that is used for all new files created in that directory, and also copied to new subdirectories created therein
• the user may specify a layout specifically for that file using lfs setstripe ... <filename> that is used preferentially, though it will inherit parameters from the parent/root directory or filesystem-wide defaults if some parameter is not specified (e.g. stripe size, pool, etc)

[Andreas Dilger]

Closing this ticket as questions have been answered.