pveceph: improve intro

[pve-docs.git] / pvesr.adoc

ifdef::manvolnum[]
pvesr(1)
========
:pve-toplevel:

NAME
----

pvesr - Proxmox VE Storage Replication

SYNOPSIS
--------

include::pvesr.1-synopsis.adoc[]

DESCRIPTION
-----------
endif::manvolnum[]

ifndef::manvolnum[]
Storage Replication
===================
:pve-toplevel:
endif::manvolnum[]

The `pvesr` command line tool manages the {PVE} storage replication
framework. Storage replication brings redundancy for guests using
local storage's and reduces migration time when you migrate a guest.

It replicates virtual disks to another node so that all data is
available without using shared storage. Replication uses storage
snapshots to minimize traffic sent over the network. New data is sent
incrementally after the initial sync. In the case of a node failure,
your guest data is still available on the replicated node.

The minimal replication interval are 1 minute and the maximal interval is once a week.
Interval schedule format is a subset of `systemd` calendar events.
Every interval time your guest vdisk data will be synchronized,
but only the new data will replicated. This reduce the amount of data to a minimum.
New data are data what are written to the vdisk after the last replication.

Every guest can replicate to many target nodes, but only one replication job per target node is allowed.

The migration of guests, where storage replication is activated, is currently only offline possible.
When the guest will migrate to the target of the replication, only the delta of the data must migrated and
the replication direction will switched automatically in the opposite direction.
If you migrate to a node where you do not replicate, it will send the whole vdisk data to the new node and after the migration it continuous the replication job as usually.

WARNING: High-Availability is possible with Storage Replication but this can lead to lose data. So be aware of this problem before you use this combination.

Supported Storage Types
-----------------------

.Storage Types
[width="100%",options="header"]
|============================================
|Description    |PVE type    |Snapshots|Stable
|ZFS (local)    |zfspool     |yes      |yes
|============================================

Schedule
--------

Proxmox VE has a very flexible replication scheduler with will explained in detail here.

A schedule string has following format.

[day of the week] <time>

Day of the week are taken of sun, mon, tue, wed, thu, fri and sat.
It is allowed to make a list of this token.
Or alternative you can use a span.

The second part are the time, here the string contains two sub-string separate by `:`.
First comes the hours, than the minutes.
0..23 are the allowed values for hours and 0..59 for minutes.  
If hours is missing it will substitute with `*`, this also counts for the day of the week.
The `*` will be interpreted as wildcard. In case of the "all day of the week" it would be sun..sat, for the hours 0..23 or for the minutes 0..59.

Examples:
~~~~~~~~~

.Schedule Examples
[width="100%",options="header"]
|============================================
|String					|Meaning
|"mon,tue,wed,thu,fri" => "mon..fir"	|All working days at 0:00.
|"sat,sun" =>  "sat..sun" 		|Only on weekend at 0:00.
|"mon,wed,fri" 				|Only on Monday, Wednesday and Friday at 0:00.
|"12:05" => "sun..sat 12:05"		|All weekdays at 12:05 PM.
|"*/5"	   	     			|All five minutes on all weekdays.
|"mon..wed 30/10"			|Monday, Tuesday, Wednesday at all hours and 30, 40, 50 minutes.
|"fri 12..13:5/20"			|Friday at 12:05, 12:25, 12:45, 13:05, 13:25 and 13:45.
|"12/2:05" 				|All weekdays every 2 hours start at 12:05 until 22:05
|============================================
The simplest case is the `*` which means every min what is the minimum interval.


Error State
-----------

It can happen that a job will have a error.
Common issues for example:

Network is not working.

No space on the disk.
...

In this case storage replication will retry every 30 miuntes.


Migrating a guest in case of Error  
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Assumed that you have two guests (KVM 100| LXC 200) running on node A (pve1) and replicate to node B (pve2).
Node A is fails and can not get back online. Now you have to migrate the guest to Node B manually.

Connect to node B with ssh. 

- check if that the cluster has quorum.

 pvecm status

Is quorum is ok continuing with the skip[ the next step.

- If you have no quorum, it is important to make your pmxcfs writable.

WARNING: If expected is set, please avoid large changes on the nodes before the cluster is quorate online because there is a little change to a split brain. 

 pvecm expected 1

- move the guest config form the origin node A to node B.
 cd /etc/pve/node && mv pve1/qemu-server/100.conf pve2/qemu-server/100.conf
 cd /etc/pve/node && mv pve1/lxc/200.conf pve2/lxc/100.conf

- Now you can start the guests.

 qm start 100
 pct start 100


Create Replication
------------------

[thumbnail="gui-create-replica-on-node.png"]

You can easy create a replication job on the GUI or as well on the command line.
There are only a few setting to configure.

The only required parameter on the GUI is the target node.
If you create a job on the command line also a unique id is needed.
This id will automatically generated in case of GUI created jobs.
The id contains the VMID and a consecutive number separated by a minus.
All other parameters are optional.

Update Replication Job
----------------------

For now is possible to update rate limit and the scheduler.
It can be done on the GUI or on the cli.


Using the Command Line Interface
--------------------------------

Examples
~~~~~~~~

Create a replication job for VM 100 first job, which will run all 5 min with max bandwidth of 10MByte/s.

 pvesr create-local-job 100-0 pve1 --schedule */5 --rate 10

Disable a running job for ID 100-0 

 pvesr disable 100-0

Enable a running job for ID 100-0

 pvesr disable 100-0

Change the schedule interval to once a hour

 pvesr update 100-0 --schedule '*/00'

ifdef::manvolnum[]
include::pve-copyright.adoc[]
endif::manvolnum[]