README

System Storage Manager
**********************

A single tool to manage your storage.

This document in HTML form also at:

https://system-storage-manager.github.io/


Description
***********

System Storage Manager provides an easy to use command line interface
to manage your storage using various technologies like lvm, btrfs,
encrypted volumes and more.

In more sophisticated enterprise storage environments, management with
Device Mapper (dm), Logical Volume Manager (LVM), or Multiple Devices
(md) is becoming increasingly more difficult.  With file systems added
to the mix, the number of tools needed to configure and manage storage
has grown so large that it is simply not user friendly.  With so many
options for a system administrator to consider, the opportunity for
errors and problems is large.

The btrfs administration tools have shown us that storage management
can be simplified, and we are working to bring that ease of use to
Linux filesystems in general.


Licence
*******

(C)2017 Red Hat, Inc., Jan Tulak <jtulak@redhat.com> (C)2011 Red Hat,
Inc., Lukas Czerner <lczerner@redhat.com>

This program is free software: you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation, either version 2 of the License, or (at
your option) any later version.

This program is distributed in the hope that it will be useful, but
WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
General Public License for more details.

You should have received a copy of the GNU General Public License
along with this program.  If not, see <http://www.gnu.org/licenses/>.


Commands
********


Introduction
************

System Storage Manager has several commands that you can specify on
the command line as a first argument to ssm. They all have a specific
use and their own arguments, but global ssm arguments are propagated
to all commands.


Create command
**************

This command creates a new volume with defined parameters. If a
**device** is provided it will be used to create the volume, hence it
will be added into the **pool** prior to volume creation (See *Add
command section*). More than one device can be used to create a
volume.

If the **device** is already being used in a different pool, then
**ssm** will ask you whether you want to remove it from the original
pool. If you decline, or the removal fails, then the **volume**
creation fails if the *SIZE* was not provided. On the other hand, if
the *SIZE* is provided and some devices can not be added to the
**pool**, the volume creation might still succeed if there is enough
space in the **pool**.

In addition to specifying size of the volume directly, percentage can
be specified as well. Specify **--size 70%** to indicate the volume
size to be 70% of total pool size. Additionally, percentage of the
used, or free pool space can be specified as well using keywords FREE,
or USED respectively.

The *POOL* name can be specified as well. If the pool exists, a new
volume will be created from that pool (optionally adding **device**
into the pool). However if the *POOL* does not exist, then **ssm**
will attempt to create a new pool with the provided **device**, and
then create a new volume from this pool. If the **--backend** argument
is omitted, the default **ssm** backend will be used. The default
backend is *lvm*.

**ssm** also supports creating a RAID configuration, however some
back-ends might not support all RAID levels, or may not even support
RAID at all. In this case, volume creation will fail.

If a **mount** point is provided, **ssm** will attempt to mount the
volume after it is created. However it will fail if mountable file
system is not present on the volume.

If the backend allows it (currently only supported with **lvm**
backend), **ssm** can be used to create **thinly provisioned volumes**
by specifying **--virtual-size** option. This will automatically
create a **thin pool** of a given size provided with **--size** option
and **thin volume** of a given size provided with **--virtual-size**
option and name provided with **--name** option. Virtual size can be
much bigger than available space in the **pool**.


List command
************

Lists information about all detected devices, pools, volumes and
snapshots found on the system. The **list** command can be used either
alone to list all of the information, or you can request specific
sections only.

The following sections can be specified:

{volumes | vol}
   List information about all **volumes** found in the system.

{devices | dev}
   List information about all **devices** found on the system. Some
   devices are intentionally hidden, like for example cdrom or DM/MD
   devices since those are actually listed as volumes.

{pools | pool}
   List information about all **pools** found in the system.

{filesystems | fs}
   List information about all volumes containing **filesystems** found
   in the system.

{snapshots | snap}
   List information about all **snapshots** found in the system. Note
   that some back-ends do not support snapshotting and some cannot
   distinguish snapshot from regular volumes. In this case, **ssm**
   will try to recognize the volume name in order to identify a
   **snapshot**, but if the **ssm** regular expression does not match
   the snapshot pattern, the problematic snapshot will not be
   recognized.


Info command
************

**EXPERIMENTAL** This feature is currently experimental. The output
format can change and fields can be added or removed.

Show detailed information about all detected devices, pools, volumes
and snapshots found on the system. The **info** command can be used
either alone to show all available items, or you can specify a device,
pool, or any other identifier to see information about the specific
item.


Remove command
**************

This command removes an **item** from the system. Multiple items can
be specified.  If the **item** cannot be removed for some reason, it
will be skipped.

An **item** can be any of the following:

device
   Remove a **device** from the pool. Note that this cannot be done in
   some cases where the device is being used by the pool. You can use
   the **-f** argument to *force* removal. If the device does not
   belong to any pool, it will be skipped.

pool
   Remove a **pool** from the system. This will also remove all
   volumes created from that pool.

volume
   Remove a **volume** from the system. Note that this will fail if
   the **volume** is mounted and cannot be *forced* with **-f**.


Resize command
**************

Change size of the **volume** and file system. If there is no file
system, only the **volume** itself will be resized. You can specify a
**device** to add into the **volume** pool prior the resize. Note that
the **device** will only be added into the pool if the **volume** size
is going to grow.

If the **device** is already used in a different pool, then **ssm**
will ask you whether or not you want to remove it from the original
pool.

In some cases, the file system has to be mounted in order to resize.
This will be handled by **ssm** automatically by mounting the
**volume** temporarily.

In addition to specifying new size of the volume directly, percentage
can be specified as well. Specify **--size 70%** to resize the volume
to 70% of it's original size. Additionally, percentage of the used, or
free pool space can be specified as well using keywords FREE, or USED
respectively.

Note that resizing btrfs subvolume is not supported, only the whole
file system can be resized.


Check command
*************

Check the file system consistency on the **volume**. You can specify
multiple volumes to check. If there is no file system on the
**volume**, this **volume** will be skipped.

In some cases the file system has to be mounted in order to check the
file system.  This will be handled by **ssm** automatically by
mounting the **volume** temporarily.


Snapshot command
****************

Take a snapshot of an existing **volume**. This operation will fail if
the back-end to which the **volume** belongs to does not support
snapshotting. Note that you cannot specify both *NAME* and *DEST*
since those options are mutually exclusive.

In addition to specifying new size of the volume directly, percentage
can be specified as well. Specify **--size 70%** to indicate the new
snapshot size to be 70% of the origin volume size. Additionally,
percentage of the used, or free pool space can be specified as well
using keywords FREE, or USED respectively.

In some cases the file system has to be mounted in order to take a
snapshot of the **volume**. This will be handled by **ssm**
automatically by mounting the **volume** temporarily.


Add command
***********

This command adds a **device** into the pool. By default, the
**device** will not be added if it's already a part of a different
pool, but the user will be asked whether or not to remove the device
from its pool. When multiple devices are provided, all of them are
added into the pool. If one of the devices cannot be added into the
pool for any reason, the add command will fail. If no pool is
specified, the default pool will be chosen. In the case of a non
existing pool, it will be created using the provided devices.


Mount command
*************

This command will mount the **volume** at the specified **directory**.
The **volume** can be specified in the same way as with **mount(8)**,
however in addition, one can also specify a **volume** in the format
as it appears in the **ssm list** table.

For example, instead of finding out what the device and subvolume id
of the btrfs subvolume "btrfs_pool:vol001" is in order to mount it,
one can simply call **ssm mount btrfs_pool:vol001 /mnt/test**.

One can also specify *OPTIONS* in the same way as with **mount(8)**.


Migrate command
***************

Move data from one device to another. For btrfs and lvm their
specialized utilities are used, so the data are moved in an all-or-
nothing fashion and no other operation is needed to add/remove the
devices or rebalance the pool. Devices that do not belong to a
**backend** that supports specialized device migration tools will be
migrated using **dd**.

This operation is not intended to be used for duplication, because the
process can change metadata and an access to the data may be
difficult.


Backends
********


Introduction
************

Ssm aims to create a unified user interface for various technologies
like Device Mapper (dm), Btrfs file system, Multiple Devices (md) and
possibly more. In order to do so we have a core abstraction layer in
"ssmlib/main.py". This abstraction layer should ideally know nothing
about the underlying technology, but rather comply with **device**,
**pool** and **volume** abstractions.

Various backends can be registered in "ssmlib/main.py" in order to
handle specific storage technology, implementing methods like
*create*, *snapshot*, or *remove* volumes and pools. The core will
then call these methods to manage the storage without needing to know
what lies underneath it. There are already several backends registered
in ssm.


Btrfs backend
*************

Btrfs is the file system with many advanced features including volume
management. This is the reason why btrfs is handled differently than
other *conventional* file systems in **ssm**. It is used as a volume
management back-end.

Pools, volumes and snapshots can be created with btrfs backend and
here is what it means from the btrfs point of view:

pool
   A pool is actually a btrfs file system itself, because it can be
   extended by adding more devices, or shrunk by removing devices from
   it. Subvolumes and snapshots can also be created. When the new
   btrfs pool should be created, **ssm** simply creates a btrfs file
   system, which means that every new btrfs pool has one volume of the
   same name as the pool itself which can not be removed without
   removing the entire pool. The default btrfs pool name is
   **btrfs_pool**.

   When creating a new btrfs pool, the name of the pool is used as the
   file system label. If there is an already existing btrfs file
   system in the system without a label, a btrfs pool name will be
   generated for internal use in the following format "btrfs_{device
   base name}".

   A btrfs pool is created when the **create** or **add** command is
   used with specified devices and non existing pool name.

volume
   A volume in the btrfs back-end is actually just btrfs subvolume
   with the exception of the first volume created on btrfs pool
   creation, which is the file system itself. Subvolumes can only be
   created on the btrfs file system when it is mounted, but the user
   does not have to worry about that since **ssm** will automatically
   mount the file system temporarily in order to create a new
   subvolume.

   The volume name is used as subvolume path in the btrfs file system
   and every object in this path must exist in order to create a
   volume. The volume name for internal tracking and that is visible
   to the user is generated in the format "{pool_name}:{volume name}",
   but volumes can be also referenced by its mount point.

   The btrfs volumes are only shown in the *list* output, when the
   file system is mounted, with the exception of the main btrfs volume
   - the file system itself.

   Also note that btrfs volumes and subvolumes cannot be resized. This
   is mainly limitation of the btrfs tools which currently do not work
   reliably.

   A new btrfs volume can be created with the **create** command.

snapshot
   The btrfs file system supports subvolume snapshotting, so you can
   take a snapshot of any btrfs volume in the system with **ssm**.
   However btrfs does not distinguish between subvolumes and
   snapshots, because a snapshot is actually just a subvolume with
   some blocks shared with a different subvolume. This means, that
   **ssm** is not able to directly recognize a btrfs snapshot.
   Instead, **ssm** will try to recognize a special name format of the
   btrfs volume that denotes it is a snapshot. However, if the *NAME*
   is specified when creating snapshot which does not match the
   special pattern, snapshot will not be recognized by the **ssm** and
   it will be listed as regular btrfs volume.

   A new btrfs snapshot can be created with the **snapshot** command.

device
   Btrfs does not require a special device to be created on.


Lvm backend
***********

Pools, volumes and snapshots can be created with lvm, which pretty
much match the lvm abstraction.

pool
   An lvm pool is just a *volume group* in lvm language. It means that
   it is grouping devices and new logical volumes can be created out
   of the lvm pool. The default lvm pool name is **lvm_pool**.

   An lvm pool is created when the **create** or **add** commands are
   used with specified devices and a non existing pool name.

   Alternatively a **thin pool** can be created as a result of using
   **--virtual-size** option to create **thin volume**.

volume
   An lvm volume is just a *logical volume* in lvm language. An lvm
   volume can be created with the **create** command.

snapshot
   Lvm volumes can be snapshotted as well. When a snapshot is created
   from the lvm volume, a new *snapshot* volume is created, which can
   be handled as any other lvm volume. Unlike *btrfs* lvm is able to
   distinguish snapshot from regular volume, so there is no need for a
   snapshot name to match special pattern.

device
   Lvm requires a *physical device* to be created on the device, but
   with **ssm** this is transparent for the user.


Crypt backend
*************

The crypt backend in **ssm** uses cryptsetup and dm-crypt target to
manage encrypted volumes. The crypt backend can be used as a regular
backend for creating encrypted volumes on top of regular block
devices, or even other volumes (lvm or md volumes for example). Or it
can be used to create encrypted lvm volumes right away in a single
step.

Only volumes can be created with crypt backend. This backend does not
support pooling and does not require special devices.

pool
   The crypt backend does not support pooling, and it is not possible
   to create crypt pool or add a device into a pool.

volume
   A volume in the crypt backend is the volume created by dm-crypt
   which represents the data on the original encrypted device in
   unencrypted form. The crypt backend does not support pooling, so
   only one device can be used to create crypt volume. It also does
   not support raid or any device concatenation.

   Currently two modes, or extensions are supported: luks and plain.
   Luks is used by default. For more information about the extensions,
   please see **cryptsetup** manual page.

snapshot
   The crypt backend does not support snapshotting, however if the
   encrypted volume is created on top of an lvm volume, the lvm volume
   itself can be snapshotted. The snapshot can be then opened by using
   **cryptsetup**. It is possible that this might change in the future
   so that **ssm** will be able to activate the volume directly
   without the extra step.

device
   The crypt backend does not require a special device to be created
   on.


Multipath backend
*****************

Multipath backend in **ssm** is currently limited to only gather the
information about multipath volumes in the system. You can not create
or manage multipath volumes or pools, but this functionality will be
extended in the future.


Environment variables
*********************

SSM_DEFAULT_BACKEND
   Specify which backend will be used by default. This can be
   overridden by specifying the **-b** or **--backend** argument.
   Currently only *lvm* and *btrfs* are supported.

SSM_LVM_DEFAULT_POOL
   Name of the default lvm pool to be used if the **-p** or **--pool**
   argument is omitted.

SSM_BTRFS_DEFAULT_POOL
   Name of the default btrfs pool to be used if the **-p** or
   **--pool** argument is omitted.

SSM_PREFIX_FILTER
   When this is set, **ssm** will filter out all devices, volumes and
   pools whose name does not start with this prefix. It is used mainly
   in the **ssm** test suite to make sure that we do not scramble the
   local system configuration.


Quick examples
**************

List system storage:

   # ssm list
   ----------------------------------
   Device          Total  Mount point
   ----------------------------------
   /dev/loop0    5.00 GB
   /dev/loop1    5.00 GB
   /dev/loop2    5.00 GB
   /dev/loop3    5.00 GB
   /dev/loop4    5.00 GB
   /dev/sda    149.05 GB  PARTITIONED
   /dev/sda1    19.53 GB  /
   /dev/sda2    78.12 GB
   /dev/sda3     1.95 GB  SWAP
   /dev/sda4     1.00 KB
   /dev/sda5    49.44 GB  /mnt/test
   ----------------------------------
   ------------------------------------------------------------------------------
   Volume     Pool      Volume size  FS     FS size      Free  Type   Mount point
   ------------------------------------------------------------------------------
   /dev/dm-0  dm-crypt     78.12 GB  ext4  78.12 GB  45.01 GB  crypt  /home
   /dev/sda1               19.53 GB  ext4  19.53 GB  12.67 GB  part   /
   /dev/sda5               49.44 GB  ext4  49.44 GB  29.77 GB  part   /mnt/test
   ------------------------------------------------------------------------------

Create a volume of the defined size with the defined file system. The
default back-end is set to lvm and the lvm default pool name (volume
group) is lvm_pool:

   # ssm create --fs ext4 -s 15G /dev/loop0 /dev/loop1

The name of the new volume is '/dev/lvm_pool/lvol001'. Resize the
volume to 10GB:

   # ssm resize -s-5G /dev/lvm_pool/lvol001

Resize the volume to 100G, but it may require adding more devices into
the pool:

   # ssm resize -s 100G /dev/lvm_pool/lvol001 /dev/loop2

Now we can try to create a new lvm volume named 'myvolume' from the
remaining pool space with the xfs file system and mount it to
/mnt/test1:

   # ssm create --fs xfs --name myvolume /mnt/test1

List all volumes with file systems:

   # ssm list filesystems
   -----------------------------------------------------------------------------------------------
   Volume                  Pool        Volume size  FS      FS size      Free  Type    Mount point
   -----------------------------------------------------------------------------------------------
   /dev/lvm_pool/lvol001   lvm_pool       25.00 GB  ext4   25.00 GB  23.19 GB  linear
   /dev/lvm_pool/myvolume  lvm_pool        4.99 GB  xfs     4.98 GB   4.98 GB  linear  /mnt/test1
   /dev/dm-0               dm-crypt       78.12 GB  ext4   78.12 GB  45.33 GB  crypt   /home
   /dev/sda1                              19.53 GB  ext4   19.53 GB  12.67 GB  part    /
   /dev/sda5                              49.44 GB  ext4   49.44 GB  29.77 GB  part    /mnt/test
   -----------------------------------------------------------------------------------------------

You can then easily remove the old volume with:

   # ssm remove /dev/lvm_pool/lvol001

Now let's try to create a btrfs volume. Btrfs is a separate backend,
not just a file system. That is because btrfs itself has an integrated
volume manager. The default btrfs pool name is btrfs_pool.:

   # ssm -b btrfs create /dev/loop3 /dev/loop4

Now we create btrfs subvolumes. Note that the btrfs file system has to
be mounted in order to create subvolumes. However ssm will handle this
for you.:

   # ssm create -p btrfs_pool
   # ssm create -n new_subvolume -p btrfs_pool


   # ssm list filesystems
   -----------------------------------------------------------------
   Device         Free      Used      Total  Pool        Mount point
   -----------------------------------------------------------------
   /dev/loop0  0.00 KB  10.00 GB   10.00 GB  lvm_pool
   /dev/loop1  0.00 KB  10.00 GB   10.00 GB  lvm_pool
   /dev/loop2  0.00 KB  10.00 GB   10.00 GB  lvm_pool
   /dev/loop3  8.05 GB   1.95 GB   10.00 GB  btrfs_pool
   /dev/loop4  6.54 GB   1.93 GB    8.47 GB  btrfs_pool
   /dev/sda                       149.05 GB              PARTITIONED
   /dev/sda1                       19.53 GB              /
   /dev/sda2                       78.12 GB
   /dev/sda3                        1.95 GB              SWAP
   /dev/sda4                        1.00 KB
   /dev/sda5                       49.44 GB              /mnt/test
   -----------------------------------------------------------------
   -------------------------------------------------------
   Pool        Type   Devices     Free      Used     Total
   -------------------------------------------------------
   lvm_pool    lvm    3        0.00 KB  29.99 GB  29.99 GB
   btrfs_pool  btrfs  2        3.84 MB  18.47 GB  18.47 GB
   -------------------------------------------------------
   -----------------------------------------------------------------------------------------------
   Volume                  Pool        Volume size  FS      FS size      Free  Type    Mount point
   -----------------------------------------------------------------------------------------------
   /dev/lvm_pool/lvol001   lvm_pool       25.00 GB  ext4   25.00 GB  23.19 GB  linear
   /dev/lvm_pool/myvolume  lvm_pool        4.99 GB  xfs     4.98 GB   4.98 GB  linear  /mnt/test1
   /dev/dm-0               dm-crypt       78.12 GB  ext4   78.12 GB  45.33 GB  crypt   /home
   btrfs_pool              btrfs_pool     18.47 GB  btrfs  18.47 GB  18.47 GB  btrfs
   /dev/sda1                              19.53 GB  ext4   19.53 GB  12.67 GB  part    /
   /dev/sda5                              49.44 GB  ext4   49.44 GB  29.77 GB  part    /mnt/test
   -----------------------------------------------------------------------------------------------

Now let's free up some of the loop devices so that we can try to add
them into the btrfs_pool. So we'll simply remove lvm myvolume and
resize lvol001 so we can remove /dev/loop2. Note that myvolume is
mounted so we have to unmount it first.:

   # umount /mnt/test1
   # ssm remove /dev/lvm_pool/myvolume
   # ssm resize -s-10G /dev/lvm_pool/lvol001
   # ssm remove /dev/loop2

Add device to the btrfs file system:

   # ssm add /dev/loop2 -p btrfs_pool

Now let's see what happened. Note that to actually see btrfs
subvolumes you have to mount the file system first:

   # mount -L btrfs_pool /mnt/test1/
   # ssm list volumes
   ------------------------------------------------------------------------------------------------------------------------
   Volume                         Pool        Volume size  FS      FS size      Free  Type    Mount point
   ------------------------------------------------------------------------------------------------------------------------
   /dev/lvm_pool/lvol001          lvm_pool       15.00 GB  ext4   15.00 GB  13.85 GB  linear
   /dev/dm-0                      dm-crypt       78.12 GB  ext4   78.12 GB  45.33 GB  crypt   /home
   btrfs_pool                     btrfs_pool     28.47 GB  btrfs  28.47 GB  28.47 GB  btrfs   /mnt/test1
   btrfs_pool:2012-05-09-T113426  btrfs_pool     28.47 GB  btrfs  28.47 GB  28.47 GB  btrfs   /mnt/test1/2012-05-09-T113426
   btrfs_pool:new_subvolume       btrfs_pool     28.47 GB  btrfs  28.47 GB  28.47 GB  btrfs   /mnt/test1/new_subvolume
   /dev/sda1                                     19.53 GB  ext4   19.53 GB  12.67 GB  part    /
   /dev/sda5                                     49.44 GB  ext4   49.44 GB  29.77 GB  part    /mnt/test
   ------------------------------------------------------------------------------------------------------------------------

Remove the whole lvm pool, one of the btrfs subvolumes, and one unused
device from the btrfs pool btrfs_loop3. Note that with btrfs, pools
have the same name as their volumes:

   # ssm remove lvm_pool /dev/loop2 /mnt/test1/new_subvolume/

Snapshots can also be done with ssm:

   # ssm snapshot btrfs_pool
   # ssm snapshot -n btrfs_snapshot btrfs_pool

With lvm, you can also create snapshots:

   # ssm create -s 10G /dev/loop[01]
   # ssm snapshot /dev/lvm_pool/lvol001

Now list all snapshots. Note that btrfs snapshots are actually just
subvolumes with some blocks shared with the original subvolume, so
there is currently no way to distinguish between those. ssm is using a
little trick to search for name patterns to recognize snapshots, so if
you specify your own name for the snapshot, ssm will not recognize it
as snapshot, but rather as regular volume (subvolume). This problem
does not exist with lvm.:

   # ssm list snapshots
   -------------------------------------------------------------------------------------------------------------
   Snapshot                            Origin   Volume size     Size  Type    Mount point
   -------------------------------------------------------------------------------------------------------------
   /dev/lvm_pool/snap20120509T121611   lvol001      2.00 GB  0.00 KB  linear
   btrfs_pool:snap-2012-05-09-T121313              18.47 GB           btrfs   /mnt/test1/snap-2012-05-09-T121313
   -------------------------------------------------------------------------------------------------------------


Installation
************

To install System Storage Manager into your system simply run:

   python setup.py install

as root in the System Storage Manager directory. Make sure that your
system configuration meets the *requirements* in order for ssm to work
correctly.

Note that you can run **ssm** even without installation by using the
local sources with:

   bin/ssm.local


Requirements
************

Python 2.6 or higher is required to run this tool. System Storage
Manager can only be run as root since most of the commands require
root privileges.

There are other requirements listed below, but note that you do not
necessarily need all dependencies for all backends. However if some of
the tools required by a backend are missing, that backend will not
work.


Python modules
==============

* argparse

* atexit

* base64

* datetime

* fcntl

* getpass

* os

* pwquality

* re

* socket

* stat

* struct

* subprocess

* sys

* tempfile

* termios

* threading

* tty


System tools
============

* tune2fs

* fsck.SUPPORTED_FS

* resize2fs

* xfs_db

* xfs_check

* xfs_growfs

* mkfs.SUPPORTED_FS

* which

* mount

* blkid

* wipefs

* dd


Lvm backend
===========

* lvm2 binaries

Some distributions (e.g. Debian) have thin provisioning tools for LVM
as an optional dependency, while others install it automatically. Thin
provisioning without these tools installed is not supported by SSM.


Btrfs backend
=============

* btrfs progs


Crypt backend
=============

* dmsetup

* cryptsetup


Multipath backend
=================

* multipath


For developers
**************

We are accepting patches! If you're interested in contributing to the
System Storage Manager code, just checkout the git repository located
on SourceForge. Please, base all of your work on the "devel" branch
since it is more up-to-date and it will save us some work when merging
your patches:

   git clone --branch devel git@github.com:system-storage-manager/ssm.git storagemanager-code

Any form of contribution - patches, documentation, reviews or rants
are appreciated. See *Mailing list section* section.


Tests
=====

System Storage Manager contains a regression testing suite to make
sure that we do not break things that should already work. We
recommend that every developer run these tests before sending patches:

   python test.py

Tests in System Storage Manager are divided into four levels.

1. First the doctest is executed.

2. Then we have unittests in "tests/unittests/test_ssm.py" which is
   testing the core of ssm "ssmlib/main.py". It is checking for basic
   things like required backend methods and variables, flag
   propagations, proper class initialization and finally whether
   commands actually result in the proper backend callbacks. It does
   not require root permissions and it does not touch your system
   configuration in any way. It actually should not invoke any shell
   command, and if it does it's a bug.

3. Second part of unittests is backend testing. We are mainly testing
   whether ssm commands result in proper backend operations. It does
   not require root permissions and it does not touch your system
   configuration in any way. It actually should not invoke any shell
   command and if it does it's a bug.

4. And finally there are real bash tests located in "tests/bashtests".
   Bash tests are divided into files. Each file tests one command for
   one backend and it contains a series of test cases followed by
   checks as to whether the command created the expected result. In
   order to test real system commands we have to create a system
   device to test on and not touch the existing system configuration.

   Before each test a number of devices are created using *dmsetup* in
   the test directory. These devices will be used in test cases
   instead of real devices.  Real operations are performed in those
   devices as they would be on the real system devices. This phase
   requires root privileges and it will not be run otherwise. In order
   to make sure that **ssm** does not touch any existing system
   configuration, each device, pool and volume name includes a special
   prefix, and the SSM_PREFIX_FILTER environment variable is set to
   make **ssm** to exclude all items which does not match this special
   prefix.

   The standard place for temporary files is within SSM tests
   directory. However, if for some reason you don't want/can't use
   this location, set the SSM_TEST_DIR environment variable to any
   other path.

   Even though we tried hard to make sure that the bash tests do not
   change your system configuration, we recommend you **not** to run
   tests with root privileges on your work or production system, but
   rather to run them on your testing machine.

If you change or create new functionality, please make sure that it is
covered by the System Storage Manager regression test suite to make
sure that we do not break it unintentionally.

Important: Please, make sure to run full tests before you send a patch to the
  mailing list. To do so, simply run "python test.py" as root on your
  test machine.


Documentation
=============

System Storage Manager documentation is stored in the "doc/"
directory. The documentation is built using **sphinx** software which
helps us not to duplicate text for different types of documentation
(man page, html pages, readme). If you are going to modify
documentation, please make sure not to modify manual page, html pages
or README directly, but rather modify the "doc/*.rst" and
"doc/src/*.rst" files accordingly so that the change is propagated to
all documents.

Moreover, parts of the documentation such as *synopsis* or ssm command
*options* are parsed directly from the ssm help output. This means
that when you're going to add or change arguments into **ssm** the
only thing you have to do is to add or change it in the
"ssmlib/main.py" source code and then run "make dist" in the "doc/"
directory and all the documents should be updated automatically.

Important: Please make sure you update the documentation when you add or change
  **ssm** functionality if the format of the change requires it. Then
  regenerate all the documents using "make dist" and include changes
  in the patch.


Mailing list
============

System Storage Manager developers communicate via the mailing list.
The address of our mailing list is storagemanager-
devel@lists.sourceforge.net and you can subscribe on the SourceForge
project page https://lists.sourceforge.net/lists/listinfo
/storagemanager-devel. Mailing list archives can be found here
http://sourceforge.net/mailarchive/forum.php?forum_name
=storagemanager-devel.

This is also the list where patches are sent and where the review
process is happening. We do not have a separate *user* mailing list,
so feel free to drop your questions there as well.


Posting patches
===============

As already mentioned, we are accepting patches! And we are very happy
for every contribution. If you're going to send a patch in, please
make sure to follow some simple rules:

1. Before you're going to post a patch, please run our regression
   testing suite to make sure that your change does not break someone
   else's work. See *Tests section*

2. If you're making a change that might require documentation update,
   please update the documentation as well. See *Documentation
   section*

3. Make sure your patch has all the requisites such as a *short
   description* preferably 50 characters long at max describing the
   main idea of the change. *Long description* describing what was
   changed with and why and finally Signed-off-by tag.

4. The preferred way of accepting patches is through pull requests on
   GitHub, but it is possible to send them to the mailing list if you
   don't have GitHub account.

5. If you're going to send a patch to the mailing list, please send
   the patch inlined in the email body. It is much better for review
   process.

Hint: You can use **git** to do all the work for you. "git format-patch"
  and "git send-email" will help you with creating and sending the
  patch to the mailing list.