Re: management tools?

Thomas,

Thomas Goirand wrote:
> On 12/24/2012 10:12 PM, Miles Fidelman wrote:
>> I'm getting ready to rebuild a small hosting cluster, and this time
>> around add more automation to my management approach.  Which leads me to
>> wonder: Those of you who provide hosting services using Debian, what are
>> you using for:
>> - initial o/s install and configuration (e.g., FAI, other things?)
>> - software install/update/configuration (chef, puppet, ?)
>> - virtual machine management
>> - user management (control panels, ...)
>> - overall management (nagios, webmin, ...)
>>
>> Sort of trying to get sense of what people actually use, and in what
>> combinations, rather than what's got the buzz this week.
>>
>>
>> Hi,
>>
>> It all depends what you want to do. If you want to host domain names,
>> email accounts, etc., then probably you will want to try DTC, available
>> from Debian experimental. It has support for Xen VMs too, and
>> accounting, invoicing, billing, etc.:
>> http://dtcsupport.gplhost.com/PmWiki/DebianExpressSetup
>>
>> If you are searching for cloud, then probably you should have a look here:
>> https://wiki.debian.org/OpenStackHowto/Folsom
>>
>> Unfortunately, it doesn't look nice, after the Debian wiki hack, and I
>> can't edit it, but hopefully, this will change in the near future.
>> I'm involved in both projects, so I'd be happy to help / get feedback.
>>   

First off, thanks!

I'm somewhat in between the two.  Our cluster supports a mix of our own 
production services (email lists and a web site, mostly) in one VM, and 
a bunch of VMs used for various development projects and test services.  
The production VM is set up for high-availability failover (DRBD + 
Pacemaker).  The other VMs get built up, torn down, and reconfigured all 
the time.  Right now, we have two nodes in the cluster, and two 
additional stand-alone boxes - and I'm getting ready to combine 
everything into a 4-node cluster.

A full cloud installation is overkill, and most of the stacks are 
oriented toward having separate storage and compute nodes (our boxes are 
each server + 4 large drives).  Ganetti is close to what I'm looking for 
managing nodes and VMs - but it doesn't handle automatic failover, and 
is just a bit too monolithic for me to feel comfortable overlaying 
pacemaker (looks like the control code would work at cross purposes).

So, I've been looking at alternatives for basic provisioning, 
configuration, and monitoring.  One approach is to find building blocks 
and combine them myself (hence my query).  The other is to start with 
something like Webmin or DTC, and DTC keeps popping up.

Can you say a little more about DTC as to:
- general maturity, stability, community support
- how helpful is it in keeping track of IP addresses and DNS records 
(I've found keeping track of them to be the most annoying part of 
provisioning VMs)
- how helpful is it in terms of configuring, managing, and monitoring 
disks & filesystems (the other pain in the rear when setting up and 
tearing down VMs)
- does it know anything about pacemaker?

Thanks again,

Miles Fidelman


-- 
In theory, there is no difference between theory and practice.
In practice, there is.   .... Yogi Berra