mirror of
https://github.com/trapexit/mergerfs.git
synced 2026-11-20 23:16:27 -05:00
Update docs; add nofs, policyfs and greyhole
Some checks are pending
mkdocs / deploy (push) Waiting to run
Some checks are pending
mkdocs / deploy (push) Waiting to run
This commit is contained in:
parent
06433a0548
commit
ba0584a4f0
10 changed files with 183 additions and 81 deletions
|
|
@ -58,13 +58,13 @@ read/writes. Contact the software author or worse case the mergerfs
|
||||||
author so it may be investigated further.
|
author so it may be investigated further.
|
||||||
|
|
||||||
|
|
||||||
### write benchmark
|
### Write Benchmark
|
||||||
|
|
||||||
```
|
```
|
||||||
$ dd if=/dev/zero of=/mnt/mergerfs/16GB.file bs=1M count=16384 oflag=nocache conv=fdatasync status=progress
|
$ dd if=/dev/zero of=/mnt/mergerfs/16GB.file bs=1M count=16384 oflag=nocache conv=fdatasync status=progress
|
||||||
```
|
```
|
||||||
|
|
||||||
### read benchmark
|
### Read Benchmark
|
||||||
|
|
||||||
```
|
```
|
||||||
$ dd if=/mnt/mergerfs/16GB.file of=/dev/null bs=1M iflag=nocache conv=fdatasync status=progress
|
$ dd if=/mnt/mergerfs/16GB.file of=/dev/null bs=1M iflag=nocache conv=fdatasync status=progress
|
||||||
|
|
@ -79,21 +79,21 @@ Below are some basic tests but also try setting `--direct=0`, changing
|
||||||
the `ioengine`, or playing with other options.
|
the `ioengine`, or playing with other options.
|
||||||
|
|
||||||
|
|
||||||
#### write
|
#### Write
|
||||||
|
|
||||||
```
|
```
|
||||||
$ fio --name=writetest --filename=/mnt/mergerfs/16GB.file --size=16Gb --rw=write --bs=1M --direct=1 --numjobs=1 --iodepth=8 --group_reporting --runtime=60 --startdelay=0 --ioengine=psync
|
$ fio --name=writetest --filename=/mnt/mergerfs/16GB.file --size=16Gb --rw=write --bs=1M --direct=1 --numjobs=1 --iodepth=8 --group_reporting --runtime=60 --startdelay=0 --ioengine=psync
|
||||||
```
|
```
|
||||||
|
|
||||||
|
|
||||||
#### read
|
#### Read
|
||||||
|
|
||||||
```
|
```
|
||||||
$ fio --name=readtest --filename=/mnt/mergerfs/16GB.file --size=16Gb --rw=read --bs=1M --direct=1 --numjobs=1 --iodepth=8 --group_reporting --runtime=60 --startdelay=0 --ioengine=psync
|
$ fio --name=readtest --filename=/mnt/mergerfs/16GB.file --size=16Gb --rw=read --bs=1M --direct=1 --numjobs=1 --iodepth=8 --group_reporting --runtime=60 --startdelay=0 --ioengine=psync
|
||||||
```
|
```
|
||||||
|
|
||||||
|
|
||||||
### other benchmarks
|
### Other Benchmarks
|
||||||
|
|
||||||
If you are attempting to benchmark other behaviors you must ensure you
|
If you are attempting to benchmark other behaviors you must ensure you
|
||||||
clear kernel caches before runs. In fact it would be a good deal to
|
clear kernel caches before runs. In fact it would be a good deal to
|
||||||
|
|
|
||||||
|
|
@ -90,7 +90,7 @@ filesystems. Mainly `ext4`.
|
||||||
|
|
||||||
Filesystems, and therefore mergerfs, are doing **lots** of small
|
Filesystems, and therefore mergerfs, are doing **lots** of small
|
||||||
actions at high speed. It simply isn't reasonable to log all the
|
actions at high speed. It simply isn't reasonable to log all the
|
||||||
actions of the system all time time. However, there is a debug mode
|
actions of the system all the time. However, there is a debug mode
|
||||||
which can be toggled on as needed, even at runtime, which will record
|
which can be toggled on as needed, even at runtime, which will record
|
||||||
some information to help with debugging. The main thing it will record
|
some information to help with debugging. The main thing it will record
|
||||||
is a trace of all FUSE messages to the location defined by the
|
is a trace of all FUSE messages to the location defined by the
|
||||||
|
|
|
||||||
|
|
@ -170,7 +170,7 @@ may need to disable the use of the inode in the software if possible
|
||||||
or change the [inodecalc](config/inodecalc.md) setting if possible.
|
or change the [inodecalc](config/inodecalc.md) setting if possible.
|
||||||
|
|
||||||
Generally speaking, in modern setups where there are many non-fully
|
Generally speaking, in modern setups where there are many non-fully
|
||||||
complient POSIX filesystem in use, reliance on inode to have a
|
compliant POSIX filesystems in use, reliance on inode to have a
|
||||||
specific meaning is not always reasonable. FUSE, for instance, has no
|
specific meaning is not always reasonable. FUSE, for instance, has no
|
||||||
requirement that inode values need to be unique or that two files that
|
requirement that inode values need to be unique or that two files that
|
||||||
happen to be the same underlying file have the same inode.
|
happen to be the same underlying file have the same inode.
|
||||||
|
|
|
||||||
|
|
@ -17,9 +17,11 @@ and offers better performance. As of 2020 the author of mhddfs has
|
||||||
|
|
||||||
Below is an example of mhddfs and mergerfs setup to work similarly.
|
Below is an example of mhddfs and mergerfs setup to work similarly.
|
||||||
|
|
||||||
`mhddfs -o mlimit=4G,allow_other /mnt/drive1,/mnt/drive2 /mnt/pool`
|
```
|
||||||
|
mhddfs -o mlimit=4G,allow_other /mnt/drive1,/mnt/drive2 /mnt/pool
|
||||||
|
|
||||||
`mergerfs -o minfreespace=4G,category.create=ff /mnt/drive1:/mnt/drive2 /mnt/pool`
|
mergerfs -o minfreespace=4G,category.create=ff /mnt/drive1:/mnt/drive2 /mnt/pool
|
||||||
|
```
|
||||||
|
|
||||||
|
|
||||||
## aufs
|
## aufs
|
||||||
|
|
@ -30,15 +32,15 @@ Below is an example of mhddfs and mergerfs setup to work similarly.
|
||||||
aufs, another union filesystem, is a kernel based overlay filesystem
|
aufs, another union filesystem, is a kernel based overlay filesystem
|
||||||
with basic file creation placement policies.
|
with basic file creation placement policies.
|
||||||
|
|
||||||
While aufs still is maintained it failed to be included in the
|
aufs failed to be included in the mainline kernel and is no longer
|
||||||
mainline kernel and is no longer available in most Linux distros
|
available in most Linux distros, making it harder to get installed for
|
||||||
making it harder to get installed for the average user.
|
the average user. Development has been largely dormant for years.
|
||||||
|
|
||||||
While aufs can often offer better peak performance due to being
|
While aufs can often offer better peak performance due to being
|
||||||
primarily kernel based (at least when `passthrough.io` is disabled),
|
primarily kernel based (at least when `passthrough.io` is disabled in
|
||||||
mergerfs provides more configurability and is generally easier to
|
mergerfs), mergerfs provides more configurability and is generally
|
||||||
use. mergerfs however does not offer the overlay / copy-on-write (CoW)
|
easier to use. mergerfs however does not offer the overlay /
|
||||||
features which aufs has.
|
copy-on-write (CoW) features which aufs has.
|
||||||
|
|
||||||
|
|
||||||
## unionfs
|
## unionfs
|
||||||
|
|
@ -74,14 +76,13 @@ as unionfs, aufs, and overlayfs require.
|
||||||
|
|
||||||
* [https://docs.kernel.org/filesystems/overlayfs.html](https://docs.kernel.org/filesystems/overlayfs.html)
|
* [https://docs.kernel.org/filesystems/overlayfs.html](https://docs.kernel.org/filesystems/overlayfs.html)
|
||||||
|
|
||||||
overlayfs is effectively the successor to unionfs, unionfs-fuse, and
|
overlayfs is effectively the functional successor to unionfs,
|
||||||
aufs and is widely used by Linux container platforms such as Docker and
|
unionfs-fuse, and aufs and is widely used by Linux container platforms
|
||||||
Podman. It was developed and is maintained by the same developer who
|
such as Docker and Podman. Both overlayfs and FUSE were originally created by Miklos Szeredi.
|
||||||
created FUSE.
|
|
||||||
|
|
||||||
If your use case is layering a writable filesystem on top of read-only
|
If your use case is layering a writable filesystem on top of read-only
|
||||||
filesystems then you should look first to overlayfs. Its feature set
|
filesystems then you should look first to overlayfs. Its feature set
|
||||||
however is very different from mergerfs and solve different problems.
|
however is very different from mergerfs and solves different problems.
|
||||||
|
|
||||||
|
|
||||||
## RAID0, JBOD, SPAN, drive concatenation, striping
|
## RAID0, JBOD, SPAN, drive concatenation, striping
|
||||||
|
|
@ -110,18 +111,6 @@ used those other technologies. Meaning you can't create a file greater
|
||||||
than 1TB on a pool of two 1TB filesystems.
|
than 1TB on a pool of two 1TB filesystems.
|
||||||
|
|
||||||
|
|
||||||
## BTRFS Single Data Profile
|
|
||||||
|
|
||||||
[BTRFS'](https://btrfs.readthedocs.io) `single` data profile is
|
|
||||||
similar to RAID0, spreading data across multiple devices but offering
|
|
||||||
no redundancy. Unlike mergerfs which pools existing filesystems at a
|
|
||||||
high level, BTRFS is a complete filesystem that manages storage
|
|
||||||
directly. If a single device fails in BTRFS single mode, you lose all
|
|
||||||
data. mergerfs takes a different approach: it pools filesystems as-is
|
|
||||||
without redundancy, so a device failure only affects data on that one
|
|
||||||
device, not the entire pool.
|
|
||||||
|
|
||||||
|
|
||||||
## RAID5, RAID6
|
## RAID5, RAID6
|
||||||
|
|
||||||
* [RAID5](https://en.wikipedia.org/wiki/Standard_RAID_levels#RAID_5)
|
* [RAID5](https://en.wikipedia.org/wiki/Standard_RAID_levels#RAID_5)
|
||||||
|
|
@ -134,11 +123,11 @@ data once the device is replaced.
|
||||||
|
|
||||||
mergerfs offers no parity or redundancy features so in that regard the
|
mergerfs offers no parity or redundancy features so in that regard the
|
||||||
technologies are not comparable. [SnapRAID](https://www.snapraid.it)
|
technologies are not comparable. [SnapRAID](https://www.snapraid.it)
|
||||||
or [nonraid](https://github.com/qvr/nonraid) can be used in
|
or [NonRAID](https://github.com/qvr/nonraid) can be used in
|
||||||
combination with mergerfs to provide redundancy. Unlike traditional
|
combination with mergerfs to provide redundancy. Unlike traditional
|
||||||
RAID5 or RAID6 SnapRAID works with drives of different sizes and can
|
RAID5 or RAID6 SnapRAID works with drives of different sizes and can
|
||||||
have more than 2 parity drives. However, parity calculations are not
|
have more than 2 parity drives. Parity calculations with SnapRAID are
|
||||||
done in real-time. However, nonraid is realtime.
|
not done in real-time but NonRAID's are.
|
||||||
|
|
||||||
For more details and comparison of SnapRAID to related technologies
|
For more details and comparison of SnapRAID to related technologies
|
||||||
see [https://www.snapraid.it/compare](https://www.snapraid.it/compare).
|
see [https://www.snapraid.it/compare](https://www.snapraid.it/compare).
|
||||||
|
|
@ -148,14 +137,14 @@ see [https://www.snapraid.it/compare](https://www.snapraid.it/compare).
|
||||||
|
|
||||||
* [https://unraid.net](https://unraid.net)
|
* [https://unraid.net](https://unraid.net)
|
||||||
|
|
||||||
UnRAID is a full OS and offers a (FUSE based?) filesystem which
|
UnRAID is a full OS and offers a filesystem (likely FUSE based) which
|
||||||
provides a union of filesystems like mergerfs but with the addition of
|
provides a union of filesystems like mergerfs but with the addition of
|
||||||
live parity calculation and storage. Outside parity calculations
|
live parity calculation and storage. Outside parity calculations
|
||||||
mergerfs offers more features and due to the lack of real-time parity
|
mergerfs offers more features and due to the lack of real-time parity
|
||||||
calculation can have higher peak performance. For some users mergerfs
|
calculation can have higher peak performance. For some users mergerfs
|
||||||
being open source is also preferable.
|
being open source is also preferable.
|
||||||
|
|
||||||
For semi-static data mergerfs + [SnapRAID](http://www.snapraid.it)
|
For semi-static data mergerfs + [SnapRAID](https://www.snapraid.it)
|
||||||
provides a similar, but not real-time,
|
provides a similar, but not real-time,
|
||||||
solution. [NonRAID](https://github.com/qvr/nonraid) (see below) is a
|
solution. [NonRAID](https://github.com/qvr/nonraid) (see below) is a
|
||||||
fork of UnRAID's parity calculation solution and can also be used with
|
fork of UnRAID's parity calculation solution and can also be used with
|
||||||
|
|
@ -208,8 +197,10 @@ bulk media is pooled across regular filesystems using mergerfs.
|
||||||
|
|
||||||
## ZFS AnyRAID
|
## ZFS AnyRAID
|
||||||
|
|
||||||
[ZFS
|
* [https://github.com/openzfs/zfs/discussions/16773](https://github.com/openzfs/zfs/discussions/16773)
|
||||||
AnyRAID](https://hexos.com/blog/introducing-zfs-anyraid-sponsored-by-eshtek)
|
|
||||||
|
[ZFS AnyRAID](https://github.com/openzfs/zfs/discussions/16773)
|
||||||
|
([announcement](https://hexos.com/blog/introducing-zfs-anyraid-sponsored-by-eshtek))
|
||||||
is a feature being developed for ZFS to provide more flexibility in
|
is a feature being developed for ZFS to provide more flexibility in
|
||||||
pools by allowing mixed-capacity disks while maintaining live
|
pools by allowing mixed-capacity disks while maintaining live
|
||||||
redundancy.
|
redundancy.
|
||||||
|
|
@ -220,7 +211,7 @@ already provides this flexibility today: add drives of any size at any
|
||||||
time with no redundancy overhead. If you need redundancy with that
|
time with no redundancy overhead. If you need redundancy with that
|
||||||
flexibility, ZFS AnyRAID could be an option when available; until
|
flexibility, ZFS AnyRAID could be an option when available; until
|
||||||
then, mergerfs remains the simpler choice for mixed-capacity pooling
|
then, mergerfs remains the simpler choice for mixed-capacity pooling
|
||||||
with redendancy and integrity available via SnapRAID and/or NonRAID.
|
with redundancy and integrity available via SnapRAID and/or NonRAID.
|
||||||
|
|
||||||
|
|
||||||
## Bcachefs
|
## Bcachefs
|
||||||
|
|
@ -258,10 +249,26 @@ a pooling layer, providing built-in redundancy and automatic data
|
||||||
placement rather than relying on external tools or the properties of
|
placement rather than relying on external tools or the properties of
|
||||||
underlying filesystems.
|
underlying filesystems.
|
||||||
|
|
||||||
Bcachefs is under active development and as of early 2026 should be
|
Bcachefs is under active development but has had a turbulent
|
||||||
considered beta quality. It is suitable for testing and non-critical
|
relationship with the mainline kernel: it was merged into Linux 6.7
|
||||||
deployments, but careful evaluation is recommended before use in
|
(January 2024) but subsequently removed in Linux 6.18 (late 2025)
|
||||||
production systems.
|
following disagreements over development practices. It is now
|
||||||
|
distributed as an external DKMS module. It should be considered beta
|
||||||
|
quality and evaluated carefully before use in production systems.
|
||||||
|
|
||||||
|
|
||||||
|
## Btrfs Single Data Profile
|
||||||
|
|
||||||
|
* [https://btrfs.readthedocs.io](https://btrfs.readthedocs.io)
|
||||||
|
|
||||||
|
[Btrfs'](https://btrfs.readthedocs.io) `single` data profile is
|
||||||
|
similar to RAID0, spreading data across multiple devices but offering
|
||||||
|
no redundancy. Unlike mergerfs which pools existing filesystems at a
|
||||||
|
high level, Btrfs is a complete filesystem that manages storage
|
||||||
|
directly. If a single device fails in Btrfs single mode, you lose all
|
||||||
|
data. mergerfs takes a different approach: it pools filesystems as-is
|
||||||
|
without redundancy, so a device failure only affects data on that one
|
||||||
|
device, not the entire pool.
|
||||||
|
|
||||||
|
|
||||||
## StableBit's DrivePool
|
## StableBit's DrivePool
|
||||||
|
|
@ -305,7 +312,7 @@ choice. For Linux users seeking lightweight pooling without redundancy
|
||||||
overhead, mergerfs is the better option.
|
overhead, mergerfs is the better option.
|
||||||
|
|
||||||
|
|
||||||
## Plan9 binds
|
## Plan 9 binds
|
||||||
|
|
||||||
* [https://en.wikipedia.org/wiki/Plan_9_from_Bell_Labs#Union_directories_and_namespaces](https://en.wikipedia.org/wiki/Plan_9_from_Bell_Labs#Union_directories_and_namespaces)
|
* [https://en.wikipedia.org/wiki/Plan_9_from_Bell_Labs#Union_directories_and_namespaces](https://en.wikipedia.org/wiki/Plan_9_from_Bell_Labs#Union_directories_and_namespaces)
|
||||||
|
|
||||||
|
|
@ -321,6 +328,8 @@ Plan 9 isn't a widely used OS so this comparison is mostly academic.
|
||||||
|
|
||||||
## SnapRAID pooling
|
## SnapRAID pooling
|
||||||
|
|
||||||
|
* [https://www.snapraid.it/manual](https://www.snapraid.it/manual)
|
||||||
|
|
||||||
[SnapRAID](https://www.snapraid.it/manual) has a pooling feature which
|
[SnapRAID](https://www.snapraid.it/manual) has a pooling feature which
|
||||||
creates "a read-only virtual view of all the files in your array using
|
creates "a read-only virtual view of all the files in your array using
|
||||||
symbolic links."
|
symbolic links."
|
||||||
|
|
@ -341,6 +350,8 @@ that kind of setup.
|
||||||
|
|
||||||
## rclone union
|
## rclone union
|
||||||
|
|
||||||
|
* [https://rclone.org/union](https://rclone.org/union)
|
||||||
|
|
||||||
rclone's [union](https://rclone.org/union) backend allows you to
|
rclone's [union](https://rclone.org/union) backend allows you to
|
||||||
create a union of multiple rclone backends and was inspired by
|
create a union of multiple rclone backends and was inspired by
|
||||||
[mergerfs](https://rclone.org/union/#behavior-policies). Given rclone
|
[mergerfs](https://rclone.org/union/#behavior-policies). Given rclone
|
||||||
|
|
@ -355,11 +366,11 @@ given the differing feature sets and focuses of the two projects.
|
||||||
|
|
||||||
## distributed filesystems
|
## distributed filesystems
|
||||||
|
|
||||||
* AFS
|
* [AFS](https://www.openafs.org)
|
||||||
* Ceph/CephFS
|
* [Ceph/CephFS](https://ceph.io)
|
||||||
* GlusterFS
|
* [GlusterFS](https://www.gluster.org)
|
||||||
* LizardFS
|
* [LizardFS](https://lizardfs.com)
|
||||||
* MooseFS
|
* [MooseFS](https://moosefs.com)
|
||||||
* etc.
|
* etc.
|
||||||
|
|
||||||
Distributed remote filesystems come in many forms. Some offering POSIX
|
Distributed remote filesystems come in many forms. Some offering POSIX
|
||||||
|
|
@ -370,7 +381,7 @@ filesystems with duplication.
|
||||||
|
|
||||||
These filesystems almost always require a significant amount of
|
These filesystems almost always require a significant amount of
|
||||||
compute to run well and are typically deployed on their own
|
compute to run well and are typically deployed on their own
|
||||||
hardware. Often in an "orchestrators" + "workers" configuration across
|
hardware. Often in an orchestrator + worker node configuration across
|
||||||
numerous nodes. This limits their usefulness for casual and homelab
|
numerous nodes. This limits their usefulness for casual and homelab
|
||||||
users. There could also be issues with network congestion and general
|
users. There could also be issues with network congestion and general
|
||||||
performance if using a single network and that network is slower than
|
performance if using a single network and that network is slower than
|
||||||
|
|
@ -382,8 +393,98 @@ complicated setup and more compute resources than mergerfs (while also
|
||||||
offering a different set of capabilities.)
|
offering a different set of capabilities.)
|
||||||
|
|
||||||
|
|
||||||
|
## nofs
|
||||||
|
|
||||||
|
* [https://github.com/chapmanjacobd/nofs](https://github.com/chapmanjacobd/nofs)
|
||||||
|
|
||||||
|
nofs is a lightweight userspace tool that provides mergerfs-like
|
||||||
|
functionality for combining multiple filesystems/directories into a
|
||||||
|
unified view, but does so entirely through subcommands rather than
|
||||||
|
mounting a FUSE filesystem. It takes a lot of inspiration from
|
||||||
|
mergerfs in that it supports policy-based branch selection (mfs, ff,
|
||||||
|
pfrd, epmfs, and others), read-only and no-create branch modes, and
|
||||||
|
recreates POSIX-like commands such as `ls`, `find`, `which`, `cp`,
|
||||||
|
`mv`, and `rm`.
|
||||||
|
|
||||||
|
The fundamental difference from mergerfs is that nofs does not mount a
|
||||||
|
filesystem. Instead of applications transparently accessing a unified
|
||||||
|
view, users interact with the pool explicitly via nofs subcommands.
|
||||||
|
This makes it well-suited for scripting and automation, but means
|
||||||
|
applications cannot use it as a standard mountpoint and therefore 3rd
|
||||||
|
party applications won't be able to see the pool. mergerfs provides a
|
||||||
|
transparent FUSE mountpoint which applications access using normal
|
||||||
|
POSIX filesystem semantics without any modification.
|
||||||
|
|
||||||
|
nofs is a good fit if you want policy-aware file operations and branch
|
||||||
|
management from scripts or the command line and do not need a
|
||||||
|
persistent mount. mergerfs is the better choice when you need
|
||||||
|
applications to see a unified filesystem without any changes to how
|
||||||
|
they access files.
|
||||||
|
|
||||||
|
|
||||||
|
## policyfs
|
||||||
|
|
||||||
|
* [https://policyfs.org](https://policyfs.org)
|
||||||
|
* [https://github.com/hieutdo/policyfs](https://github.com/hieutdo/policyfs)
|
||||||
|
|
||||||
|
policyfs is a Linux FUSE storage daemon that unifies multiple storage
|
||||||
|
paths under a single mountpoint, similarly to mergerfs. Its
|
||||||
|
distinguishing features are explicit path-pattern-based routing rules
|
||||||
|
for reads and writes, and an optional SQLite metadata index that can
|
||||||
|
serve `readdir` and `getattr` operations without spinning up HDDs.
|
||||||
|
|
||||||
|
Where mergerfs applies a single policy across all operations (with
|
||||||
|
per-function overrides), policyfs lets you write explicit routing
|
||||||
|
rules matched against path patterns, giving finer-grained control over
|
||||||
|
which physical storage a particular directory subtree reads from or
|
||||||
|
writes to. The SQLite metadata index is a notable addition for
|
||||||
|
HDD-heavy setups where keeping disks spun down is a priority (though
|
||||||
|
how effective it is will depend on access patterns); mergerfs does not
|
||||||
|
have an equivalent cache (though does leverage kernel caches when
|
||||||
|
enabled). policyfs also supports deferred physical operations,
|
||||||
|
recording delete and rename events for indexed paths and applying them
|
||||||
|
later.
|
||||||
|
|
||||||
|
policyfs is a young project (first released in late 2025) and should
|
||||||
|
be evaluated accordingly for production use. mergerfs has a longer
|
||||||
|
track record, a broader policy set, and wider community adoption. If
|
||||||
|
spin-down-friendly metadata serving or explicit per-path routing rules
|
||||||
|
are important to your workload, policyfs is worth evaluating alongside
|
||||||
|
mergerfs.
|
||||||
|
|
||||||
|
|
||||||
|
## Greyhole
|
||||||
|
|
||||||
|
* [https://www.greyhole.net](https://www.greyhole.net)
|
||||||
|
* [https://github.com/gboudreau/Greyhole](https://github.com/gboudreau/Greyhole)
|
||||||
|
|
||||||
|
Greyhole is an open-source storage pooling application that works via
|
||||||
|
Samba. Rather than implementing a FUSE filesystem, it hooks into Samba
|
||||||
|
VFS to intercept file operations and distribute files across multiple
|
||||||
|
drives. Its headline feature compared to mergerfs is per-share
|
||||||
|
redundancy: you can configure how many copies of each file Greyhole
|
||||||
|
keeps, with copies spread across different physical drives to protect
|
||||||
|
against drive failure.
|
||||||
|
|
||||||
|
Because Greyhole operates through Samba, all file access must go
|
||||||
|
through Samba shares; it cannot be used as a general-purpose local
|
||||||
|
filesystem mount. mergerfs is a FUSE filesystem and works with any
|
||||||
|
application that uses normal POSIX filesystem calls. Greyhole is also
|
||||||
|
not well-suited for large numbers of small files or frequently-changing
|
||||||
|
files, and write performance is lower because files first land in a
|
||||||
|
temporary "landing zone" before being moved to the pool.
|
||||||
|
|
||||||
|
For users who need storage pooling with built-in per-file redundancy
|
||||||
|
and are already using Samba for file sharing, Greyhole provides both
|
||||||
|
in a single tool. For local filesystem pooling without Samba,
|
||||||
|
mergerfs is the more appropriate choice; redundancy can then be added
|
||||||
|
separately via SnapRAID or NonRAID.
|
||||||
|
|
||||||
|
|
||||||
## 9P
|
## 9P
|
||||||
|
|
||||||
|
* [https://en.wikipedia.org/wiki/9P_(protocol)](https://en.wikipedia.org/wiki/9P_(protocol))
|
||||||
|
|
||||||
[9P](https://en.wikipedia.org/wiki/9P_(protocol)) is a filesystem
|
[9P](https://en.wikipedia.org/wiki/9P_(protocol)) is a filesystem
|
||||||
protocol from the Plan 9 operating system. While historically
|
protocol from the Plan 9 operating system. While historically
|
||||||
important, it's not directly relevant for users looking to pool
|
important, it's not directly relevant for users looking to pool
|
||||||
|
|
|
||||||
|
|
@ -1,4 +1,4 @@
|
||||||
# QuickStart
|
# Quick Start
|
||||||
|
|
||||||
## Install
|
## Install
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -55,7 +55,7 @@ NFS export settings:
|
||||||
other software. By setting `softerr` or `soft` the NFS client will
|
other software. By setting `softerr` or `soft` the NFS client will
|
||||||
timeout eventually and return an error.
|
timeout eventually and return an error.
|
||||||
* `softreval`: NFS client will serve up cached data after `retrans`
|
* `softreval`: NFS client will serve up cached data after `retrans`
|
||||||
attempts to revalidate the data. Helps with intermitent network
|
attempts to revalidate the data. Helps with intermittent network
|
||||||
issues.
|
issues.
|
||||||
* `timeo=150`: Timeout till retrying request.
|
* `timeo=150`: Timeout till retrying request.
|
||||||
* `retrans=3`: Number of retrying a request.
|
* `retrans=3`: Number of retrying a request.
|
||||||
|
|
@ -122,7 +122,7 @@ line tool `uuid` or `uuidgen` or through a website such as
|
||||||
|
|
||||||
`no_root_squash` is required for the same reason mergerfs needs to run
|
`no_root_squash` is required for the same reason mergerfs needs to run
|
||||||
as `root`. Certain behaviors of mergerfs require control over the
|
as `root`. Certain behaviors of mergerfs require control over the
|
||||||
filesystem that only `root` can preform. If squashing is enabled, or
|
filesystem that only `root` can perform. If squashing is enabled, or
|
||||||
mergerfs was running as non-root, it would be unable to perform
|
mergerfs was running as non-root, it would be unable to perform
|
||||||
certain function and you will receive permission errors.
|
certain function and you will receive permission errors.
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -23,8 +23,9 @@
|
||||||
value to -990.
|
value to -990.
|
||||||
* mergerfs increases [its available file descriptor and file size
|
* mergerfs increases [its available file descriptor and file size
|
||||||
limit.](https://www.man7.org/linux/man-pages/man3/setrlimit.3p.html)
|
limit.](https://www.man7.org/linux/man-pages/man3/setrlimit.3p.html)
|
||||||
* mergerfs lowers its [scheduling
|
* mergerfs sets its [scheduling
|
||||||
priority](https://man7.org/linux/man-pages/man3/setpriority.3p.html)
|
priority](https://man7.org/linux/man-pages/man3/setpriority.3p.html)
|
||||||
to -10 ([by default](config/options.md))
|
(nice value) to -10 ([by default](config/options.md)), raising its
|
||||||
|
CPU scheduling priority above the default of 0.
|
||||||
* The [readahead](config/readahead.md) values of mergerfs itself and
|
* The [readahead](config/readahead.md) values of mergerfs itself and
|
||||||
managed filesystems can be modified.
|
managed filesystems can be modified.
|
||||||
|
|
|
||||||
|
|
@ -35,7 +35,7 @@ configure the mounting of mergerfs (/etc/fstab, systemd, etc.).
|
||||||
|
|
||||||
### Command Line Tooling
|
### Command Line Tooling
|
||||||
|
|
||||||
Extended attributes is prevelant enough that there are common tools
|
Extended attributes are prevalent enough that there are common tools
|
||||||
available for interacting with them.
|
available for interacting with them.
|
||||||
|
|
||||||
In Debian / Ubuntu distributions you can get the tools
|
In Debian / Ubuntu distributions you can get the tools
|
||||||
|
|
@ -122,14 +122,14 @@ user.mergerfs.category.search="mfs"
|
||||||
[trapexit:/mnt/mergerfs] $ getfattr -n user.mergerfs.category.create .mergerfs
|
[trapexit:/mnt/mergerfs] $ getfattr -n user.mergerfs.category.create .mergerfs
|
||||||
user.mergerfs.category.search="prfd"
|
user.mergerfs.category.search="prfd"
|
||||||
|
|
||||||
[trapexit:/mnt/mergerfs] $ setfattr -n user.mergerfs.branches -v "'+</mnt/c=RO .mergerfs
|
[trapexit:/mnt/mergerfs] $ setfattr -n user.mergerfs.branches -v '+</mnt/c=RO' .mergerfs
|
||||||
[trapexit:/mnt/mergerfs] $ getfattr -n user.mergerfs.branches .mergerfs
|
[trapexit:/mnt/mergerfs] $ getfattr -n user.mergerfs.branches .mergerfs
|
||||||
user.mergerfs.branches="/mnt/c=RO:/mnt/a=RW:/mnt/b=RW"
|
user.mergerfs.branches="/mnt/c=RO:/mnt/a=RW:/mnt/b=RW"
|
||||||
```
|
```
|
||||||
|
|
||||||
### Commands
|
### Commands
|
||||||
|
|
||||||
There are a number of commands / behaviors which can be triggerd by
|
There are a number of commands / behaviors which can be triggered by
|
||||||
writing ([setfattr](https://linux.die.net/man/1/setfattr),
|
writing ([setfattr](https://linux.die.net/man/1/setfattr),
|
||||||
[setxattr](https://linux.die.net/man/2/setxattr)) particular xattr
|
[setxattr](https://linux.die.net/man/2/setxattr)) particular xattr
|
||||||
keys of `/mountpoint/.mergerfs`. These keys do not show up in key
|
keys of `/mountpoint/.mergerfs`. These keys do not show up in key
|
||||||
|
|
|
||||||
|
|
@ -75,7 +75,7 @@ the read/write workflow, such as
|
||||||
[moveonenospc](config/moveonenospc.md), will no longer work.
|
[moveonenospc](config/moveonenospc.md), will no longer work.
|
||||||
|
|
||||||
Also, understand that this will only work on dynamically linked
|
Also, understand that this will only work on dynamically linked
|
||||||
software (a that dynamically linked with the same general libc version
|
software (that is dynamically linked with the same general libc version
|
||||||
as the software being used with it.) Anything statically compiled will
|
as the software being used with it.) Anything statically compiled will
|
||||||
not work. Many GoLang and Rust apps are statically compiled.
|
not work. Many GoLang and Rust apps are statically compiled.
|
||||||
|
|
||||||
|
|
@ -83,8 +83,8 @@ The library will not interfere with non-mergerfs filesystems. The
|
||||||
library is written to always fallback to returning the mergerfs opened
|
library is written to always fallback to returning the mergerfs opened
|
||||||
file on error.
|
file on error.
|
||||||
|
|
||||||
While the library was written to account for a number of edgecases
|
While the library was written to account for a number of edge cases
|
||||||
there could be some yet accounted for so please report any oddities.
|
there could be some yet unaccounted for so please report any oddities.
|
||||||
|
|
||||||
Thank you to
|
Thank you to
|
||||||
[nohajc](https://github.com/nohajc/mergerfs-io-passthrough) for
|
[nohajc](https://github.com/nohajc/mergerfs-io-passthrough) for
|
||||||
|
|
@ -112,7 +112,7 @@ the same path as found on the host so the preload library can see
|
||||||
them.
|
them.
|
||||||
|
|
||||||
**NOTE:** Since a container can have its own OS setup there is no
|
**NOTE:** Since a container can have its own OS setup there is no
|
||||||
guarentee that `preload.so` from the host install will be compatible
|
guarantee that `preload.so` from the host install will be compatible
|
||||||
with the loader found in the container. If that is true it simply
|
with the loader found in the container. If that is true it simply
|
||||||
won't work and shouldn't cause any issues.
|
won't work and shouldn't cause any issues.
|
||||||
|
|
||||||
|
|
|
||||||
|
|
@ -63,8 +63,8 @@ nav:
|
||||||
- setup/installation.md
|
- setup/installation.md
|
||||||
- setup/upgrade.md
|
- setup/upgrade.md
|
||||||
- setup/build.md
|
- setup/build.md
|
||||||
- terminology.md
|
- Terminology: terminology.md
|
||||||
- intro_to_filesystems.md
|
- Intro to Filesystems: intro_to_filesystems.md
|
||||||
- Config:
|
- Config:
|
||||||
- config/options.md
|
- config/options.md
|
||||||
- config/deprecated_options.md
|
- config/deprecated_options.md
|
||||||
|
|
@ -96,17 +96,17 @@ nav:
|
||||||
- config/export-support.md
|
- config/export-support.md
|
||||||
- config/kernel-permissions-check.md
|
- config/kernel-permissions-check.md
|
||||||
- config/lazy-umount-mountpoint.md
|
- config/lazy-umount-mountpoint.md
|
||||||
- error_handling_and_logging.md
|
- Error Handling and Logging: error_handling_and_logging.md
|
||||||
- resource_usage.md
|
- Resource Usage: resource_usage.md
|
||||||
- runtime_interface.md
|
- Runtime Interface: runtime_interface.md
|
||||||
- remote_filesystems.md
|
- Remote Filesystems: remote_filesystems.md
|
||||||
- tips_notes.md
|
- Tips and Notes: tips_notes.md
|
||||||
- known_issues_bugs.md
|
- Known Issues and Bugs: known_issues_bugs.md
|
||||||
- project_comparisons.md
|
- Project Comparisons: project_comparisons.md
|
||||||
- benchmarking.md
|
- Benchmarking: benchmarking.md
|
||||||
- performance.md
|
- Performance: performance.md
|
||||||
- tooling.md
|
- Tooling: tooling.md
|
||||||
- extended_usage_patterns.md
|
- Extended Usage Patterns: extended_usage_patterns.md
|
||||||
- FAQ:
|
- FAQ:
|
||||||
- faq/why_isnt_it_working.md
|
- faq/why_isnt_it_working.md
|
||||||
- faq/configuration_and_policies.md
|
- faq/configuration_and_policies.md
|
||||||
|
|
@ -116,7 +116,7 @@ nav:
|
||||||
- faq/technical_behavior_and_limitations.md
|
- faq/technical_behavior_and_limitations.md
|
||||||
- faq/have_you_considered.md
|
- faq/have_you_considered.md
|
||||||
- faq/limit_drive_spinup.md
|
- faq/limit_drive_spinup.md
|
||||||
- related_projects.md
|
- Related Projects: related_projects.md
|
||||||
- media_and_publicity.md
|
- Media and Publicity: media_and_publicity.md
|
||||||
- support.md
|
- Support: support.md
|
||||||
- sponsorship_and_donations.md
|
- Sponsorship and Donations: sponsorship_and_donations.md
|
||||||
|
|
|
||||||
Loading…
Add table
Add a link
Reference in a new issue