Raymond P. Burkholder - Things I Do - ZFS

Ramping Up For a New ZFS Project

nospam@example.com (Raymond P. Burkholder) — Sun, 10 Nov 2019 18:52:51 +0000

ZFS
OpenZFS Developer Summit 2019
Storage Configurator - from the conference
iX Systems - storage vendor
ZFS: Fun with ZFS – is compression and deduplication useful for my data and how much memory do I need for zfs dedup?
How-To: Using ZFS Encryption at Rest in OpenZFS (ZFS on Linux, ZFS on FreeBSD, …)
Performance tuning - OpenZFS

ZFS Install Notes for use with Sheepdog

nospam@example.com (Raymond P. Burkholder) — Mon, 06 Nov 2017 13:20:51 +0000

In my Sheepdog cluster, I have three nodes, with each node having two 1TB SSDs dedicated to the use of a ZFS file system. Each node stripes the two drives together to gain some read performance, and then Sheepdog will apply an Eraser Code redundancy scheme across the three nodes to provide a 2:1 erasure coded tolerant set (aka in this case similar to RAID5), which should yield about 4TB of useful storage space.

Creating the ZFS file system is a two step process: create a simple zpool, then apply the file system. This example uses two partitions on the same drive to prove the concept, but in real use, two whole drives should be used.

# step 1

# zpool create -o ashift=12 \
      -O atime=off -O canmount=off -O compression=lz4 \
      sheepdog /dev/sda7 /dev/sda8

# step 2 (this mount point is the location where sheepdog will apply its 'dog cluster format' instruction.


# zfs create sheepdog/data -o mountpoint=/var/lib/sheepdog

Followed by confirming what was defined:

# zpool status
  pool: sheepdog
 state: ONLINE
  scan: none requested
config:

        NAME        STATE     READ WRITE CKSUM
        sheepdog    ONLINE       0     0     0
          sda7      ONLINE       0     0     0
          sda8      ONLINE       0     0     0

errors: No known data errors

# zpool list
NAME       SIZE  ALLOC   FREE  EXPANDSZ   FRAG    CAP  DEDUP  HEALTH  ALTROOT
sheepdog  7.56G   444K  7.56G         -     0%     0%  1.00x  ONLINE  -
root@sw02.d01.bm1:/home/rburkholder# zfs list
NAME            USED  AVAIL  REFER  MOUNTPOINT
sheepdog        408K  7.33G    96K  /sheepdog
sheepdog/data    96K  7.33G    96K  /var/lib/sheepdog

Oracles docs on properties
Basic Notes
Creation Examples
Detailed Info
ZIL Performance: How I Doubled Sync Write Speed: background information on how the ZIL works.
OpenZFS Developer Summit 2017: talks and papers and videos from the 2017 Summit.
How to Setup ZFS Filesystem on Linux with zpool Command Examples
2018/06/24: broad overview of how ZFS is structured on disk
2018/06/24: ZFSDirectoriesAndChanges

Building ZFS on Debian Stretch

nospam@example.com (Raymond P. Burkholder) — Tue, 31 Oct 2017 12:38:29 +0000

Due to various licensing compatibility issues, which are described at What does it mean that ZFS is in Debian and On ZFS on Debian, source-only packages are available for ZFS on Debian Linux. Binaries need to be 'self-built'. Here is my method for building those binaries as packages.

I found some background information for building the packages in Debian bug #554843.

To start, add 'contrib' to /etc/apt/sources.list and run 'apt update'.

There are two dkms modules which need building: the ZFS kernel module, which depends upon the Solaris Porting Layer kernel module.

This process will need to be performed each time the kernel package gets updated or any of the related ZFS packages are updated. This process builds the kernel modules, and could be performed on a 'build machine', as various extra packages get installed to support the process:

apt install linux-headers-$(uname -r)
apt install dpkg-dev fakeroot debhelper
DEBIAN_FRONTEND=noninteractive apt-get -y --no-install-recommends install  spl-dkms
DEBIAN_FRONTEND=noninteractive apt-get -y --no-install-recommends install  zfsutils-linux zfs-zed  zfs-dkms

Packages can then be built and transported for installation on other machines:

dkms mkbmdeb spl -v 0.6.5.9 --dkmsframework framework.conf --binaries-only
dkms mkbmdeb zfs -v 0.6.5.9 --dkmsframework framework.conf --binaries-only

No Excuse for not using ZFS on Debian

nospam@example.com (Raymond P. Burkholder) — Tue, 08 Aug 2017 17:36:56 +0000

With all the BTRFS bashing going on, even though it is recommended one cares about checksummed data and metadata, there wasn't an easy alternative. That has now been solved.

ZFS is now (ie, for some time now) available as a set of native (contrib) packages in Debian. I will need to give that a test now. [zfs-zed, zfsutils-linux, zfs-dkms, zfs-initramfs, zfsutils, zfs-dracut]. With the right set of packages and boot configuration, it is also possible to use zfs as a boot partition.

Package is tracker is zfs-linux.

Primary web site is zfsonlinux.org. Documentation, FAQ, and a wiki can be found there.

A quicky setup and go can be found at HowtoForge.

NexentaStor Root / Expert Mode

nospam@example.com (Raymond P. Burkholder) — Mon, 30 Apr 2012 18:04:29 +0000

Nexenta has a pretty good web interface on their SAN product, and when that isn't good enough, there is always their web console mode. But when even that isn't good enough, and you just have to see what is happening under the hood, there is something called expert mode.

Logging in as admin, provides some good commands when dealing with the various file shares. When I changed into root mode, I've always wondered why it was so lacking. By reading some one else's site, I now know why. Nexenta has disabled most of the root stuff, and you can only access it be going a secret route:


option expert_mode=1
!bash

When finished use exit and then run their resync command to make everything right.