Backing up sakai3

[daniel_parry]

It's looking like backing up a sakai3 instance completely will require
backing up an instance that is not currently running (As part of the
jack rabbit search index is potentially stored in memory -
http://comments.gmane.org/gmane.comp.apache.jackrabbit.user/16038).
Our plan is to run two nodes in a cluster and use one of the nodes for
backup purposes only. (I.e. shut it down, back it up, bring it up
again). That way one node can remain live throughout. Was wondering
whether any one has a different / better plan?

Best wishes,

Daniel

[Ian Boston]

AFAIK thats the best way of doing it.
Ian

> --
> You received this message because you are subscribed to the Google Groups "Sakai Nakamura" group.
> To post to this group, send email to sakai-...@googlegroups.com.
> To unsubscribe from this group, send email to sakai-kernel...@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/sakai-kernel?hl=en.
>

[daniel_parry]

On 21 Sep, 11:58, Ian Boston <i...@tfd.co.uk> wrote:
> AFAIK thats the best way of doing it.

Super, thanks.
Also, which of the following files / directories need backing up (At
the moment, I assume jackrabbit only, but the activemq-data and cookie-
keystore.bin looked plausible candidates?):

activemq-data
config
cookie-keystore.bin
felix
jackrabbit
logs
nakamura.properties
org.apache.sling.launchpad.base.jar
sling.properties
startup

[Ian Boston]

No, its all folders under sling.
Ian

[daniel_parry]

On 22 Sep, 15:51, Ian Boston <i...@tfd.co.uk> wrote:
> No, its all folders under sling.

Ok, thanks. Also, if I was to use the backup of one node to seed the
start up of another to aid with faster catch up, which of those
files / directories would I put in place in the sling folder on the
new node prior to start up. Again, I would presume it was just the
jackrabbit folder?

Best wishes,

Daniel

[Carl Hall]

With nodes operating in a cluster, you will need to have shared folders between the nodes [1].  With the shared folders (and shared database), you should be able to start a clean node without pain.  If you manage your Sling/OSGi setting through configuration files, you won't need to carry over any files from/for Sling.  The search index is not shared between machines and lives in the jackrabbit folder.  IIUC, you /should/ be able to copy just the jackrabbit folder.

1 http://wiki.apache.org/jackrabbit/Clustering

[Ian Boston]

As far as Jackrabbit is concerned
Yes the jackrabbit folder.

You should make certain that there is no record in the LOCAL_REVISION table for the server in question, then when it starts up it will use the version on disk and roll the journal forwards from that point.

There might be some state in the ActiveMQ folders but I think the rest of the folders are felix state or transient.

Ian

>
> Best wishes,
>
> Daniel

[D. Stuart Freeman]

After copying the jackrabbit folder to a new node you will need to update
the repository.xml to set a unique Cluster id.

On Thu, Sep 23, 2010 at 08:56:41AM -0400, Carl Hall wrote:
> With nodes operating in a cluster, you will need to have shared folders
> between the nodes [1]. �With the shared folders (and shared database), you
> should be able to start a clean node without pain. �If you manage your
> Sling/OSGi setting through configuration files, you won't need to carry
> over any files from/for Sling. �The search index is not shared between
> machines and lives in the jackrabbit folder. �IIUC, you /should/ be able
> to copy just the jackrabbit folder.

> 1�[1]http://wiki.apache.org/jackrabbit/Clustering

> On Wed, Sep 22, 2010 at 12:09 PM, daniel_parry

> <[2]daniel...@cantab.net> wrote:

>
> On 22 Sep, 15:51, Ian Boston <[3]i...@tfd.co.uk> wrote:
> > No, its all folders under sling.
>
> Ok, thanks. Also, if I was to use the backup of one node to seed the
> start up of another to aid with faster catch up, which of those
> files / directories would I put in place in the sling folder on the
> new node prior to start up. Again, I would presume it was just the
> jackrabbit folder?
>
> Best wishes,
> Daniel
> --
> You received this message because you are subscribed to the Google
> Groups "Sakai Nakamura" group.

> To post to this group, send email to [4]sakai-...@googlegroups.com.

> To unsubscribe from this group, send email to

> [5]sakai-kernel...@googlegroups.com.

> For more options, visit this group at

> [6]http://groups.google.com/group/sakai-kernel?hl=en.

>
> --
> You received this message because you are subscribed to the Google Groups
> "Sakai Nakamura" group.
> To post to this group, send email to sakai-...@googlegroups.com.
> To unsubscribe from this group, send email to
> sakai-kernel...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/sakai-kernel?hl=en.
>

> References
>
> Visible links
> 1. http://wiki.apache.org/jackrabbit/Clustering
> 2. mailto:daniel...@cantab.net
> 3. mailto:i...@tfd.co.uk
> 4. mailto:sakai-...@googlegroups.com
> 5. mailto:sakai-kernel%2Bunsu...@googlegroups.com
> 6. http://groups.google.com/group/sakai-kernel?hl=en

--
D. Stuart Freeman
Georgia Institute of Technology
Chief Sakai 3 QA Evangelist