Hi,

Thank you so much for your response, it was very helpful.

I have a few more questions as well.
I have setup a mysql farm on scalr with 3 instances, but I am having
trouble accessing the master through a Windows instance that is also
on the same AWS account. This windows instance is hosting the domain
name I own.

The issue is that after I shut down the master mySql server (to
simulate a crash) and let a slave automatically promote itself to a
master, it takes a long time for the DNS (5 minutes at least) to be
correctly resolved on this windows instance.

If from a cmd prompt I do:
nslookup int-mysql-master.db2test.mydomain.com

I will get the follow response:
Server:  ip-172-16-0-23.ec2.internal
Address:  172.16.0.23
*** ip-172-16-0-23.ec2.internal can't find int-mysql-
master.db2test.forexonthego.com: Non-existent domain

The odd thing though, is if I keep doing the nslookup, after a while
it will finally show me the correct address, at which point I can then
login to mysql through the int-mysql-master.  If I just use the
correct internal IP that scalr shows in the dns panel, I can log in
that way (long before the dns starts resolving correctly).

Incase this helps, in Scalr the dns settings for the things I can
change are:
www 14400 IN CNAME db2test.mydomain.com.
db2test.mydomain.com. 14400 IN NS ns3.scalr.net.

At godaddy (my domain registra) I added to the name servers to have
db2test pointing to ns1.scalr.net, ns2.scalr.net, and ns3.scalr.net
Any ideas for this?

Thanks,
Cole

On Apr 1, 10:03 am, Nickolas Toursky <[email protected]> wrote:
> Hi Cole,
>
> 1. MySQL data snapshotting routine locks databases for write queries
> for some time, but they are still available for reads. Lock time
> heavily depends on data storage engine chosen. For large databases you
> should consider using either LVM or EBS storage. Anyway, exact lock
> time could only be determined experimentally for your setup.
>
> 2. Databases backup feature uses 'mysqldump' utility with all of its
> dis/advantages. Backups are stored on S3. Again, backup time depends
> on many factors and could not be predicted in general.
>
> 3. If EBS was chosen as data storage engine, all the EBS management
> will be done automatically. You don't have to setup volumes/snapshots
> on the EBS tab.
>
> 4. All the availability zones in the US are placed on the east coast.
> If you need your instances to be in the different availability zones,
> you should choose 'Place in different zones' option for 'Placement' in
> the role settings ("Placement and type" tab).
>
> 5. int-mysql subdomain round-robins through all the MySQL instances,
> this is correct. You should use int-mysql-master for write queries.
>
> Nick
>
> On Wed, Apr 1, 2009 at 8:01 AM, Cole <[email protected]> wrote:
>
> > Hello,
>
> > I've just setup Scalr so that I can get some highly available MySQL
> > servers up on EC2.  I am new, and looking for any documentation or
> > info about Scalr, and I can't hardly find anything, so if you know of
> > any, please point me in the right direction.
>
> > In the mean time, I have a few questions that would really help me out
> > if you could answer.
>
> > 1. When setting up a mysqllvm farm, if I enable "Bundle and save mysql
> > data snapshot every 48hrs"  Will the database be unavailable while the
> > bundle is taking place? If so, how long will it be unavailable?
>
> > 2. If I enable "Periodically backup databases every 15 mins" will the
> > backup cause any interruption to the database service?  If so how
> > long, and what type of backup is being performed (an export?) and
> > where is it stored?
>
> > 3. If under storage engine in the MySQL Settings tab I select EBS, do
> > I need to also specify to attach an empty EBS volume under the EBS
> > tab? Or will doing that create a second EBS drive?
>
> > 4. When I build a farm I have to select an east or west coast region.
> > How can I put a slave in a different zone for redundancy?
>
> > 5.  In the information I could find, Scalr uses a single master multi
> > slave system for MySQL -- but DNS round robins, so that write could go
> > to any of the instances, not just the master.  This seems more like a
> > master-master setup.  Which one is it?  And is there some type of
> > locking mechanism that prevents errors, such as a primary key being
> > issued to two different instances at the same time?
>
> > Thanks in advance for you help,
> > Cole

--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups 
"scalr-discuss" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at 
http://groups.google.com/group/scalr-discuss?hl=en
-~----------~----~----~----~------~----~------~--~---

Reply via email to