jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "MARTINEZ Antonio" <Antonio.Marti...@alcatel-lucent.com>
Subject RE: Snapshot of local data (in sync with DB)
Date Fri, 17 Oct 2008 02:43:56 GMT
Hi Thomas,

Thanks for your answer. I have updated the ticket.

Now, with the current implementation. What do you think is the risk of copying DB and Index
without stopping JackRabbit? (assuming JackRabbit is writing when I copy the data (DB+Index)

Do you think I will ALWAYS be able to start JackRabbit using that DB+index data (that might
not be consistent).
I do not care about having some inconsistent data, my concern is if there is a possibility
that JackRabbit would not be able to start and fix the index.

Thanks,
Antonio
 

-----Original Message-----
From: Thomas Müller [mailto:thomas.mueller@day.com] 
Sent: Monday, October 13, 2008 2:34 AM
To: users@jackrabbit.apache.org
Subject: Re: Snapshot of local data (in sync with DB)

Hi,

> Can anybody recommend a good practice to take a snapshot of the local 
> data (i.e, repository dir, revision.log file and workspaces dir), 
> guaranteeing it to be in sync with the DB (MySql in my case) at the 
> time of the snapshot?

So you are using a MySQL persistence manager, and a Lucene index? In that case you will need
to stop all cluster nodes, or at least make sure nobody writes to the repository before copying
data.

That's not a nice solution for some use cases (online backup for example). There are some
ideas how to improve it, for example using a 'Backup Manager' that would block write access
until the backup is finished.

I logged a 'feature request bug: "Support Online Backup"
https://issues.apache.org/jira/browse/JCR-1801
You could please add your use case to the bug?

Regards,
Thomas


>
> I need that for:
>
> 1) Cluster: Add a node to an existing cluster
>   I can have up to 20G DB, and indexing that in the new node would 
> take more than 3 hours, and I run out of Heap with 2G, so this is not 
> an option for me. So I would need to take a snapshot of the local data 
> in one of the cluster nodes to use it in the new node.
>
>
> 2) DB backup
>   I need to backup the DB and the local index data, but I need to 
> guarantee that they are in Sync
>
>
> 3) DB restore
>   In this case I would restore both the DB and the local data created 
> during backup
>
>
> Any help is greatly appreciated
>
> Thanks,
> Antonio
>

Mime
View raw message