Language:
switch to room list switch to menu My folders
Go to page: First ... 61 62 63 64 [65] 66 67 68 69 ... Last
[#] Sun Jul 10 2011 10:50:13 EDT from dothebart @ Uncensored

Subject: Re:

[Reply] [ReplyQuoted] [Headers] [Print]

So Jul 10 2011 10:29:23 EDT von IGnatius T Foobar @ Uncensored
Hmm. Does the stock Debian distribution have a way to do a complete in-place online upgrade to every new version, the same way Ubuntu does
?

ubuntu derived this from debian.

imho it does even work smarter in debian.

though a more frequent release of ubuntu is garanted, debian is released if its done.



[#] Sun Jul 10 2011 19:51:58 EDT from IGnatius T Foobar @ Uncensored

Subject: Re:

[Reply] [ReplyQuoted] [Headers] [Print]

Heh. And the CentOS 6.0 ISO's just landed today. :)

[#] Sun Jul 10 2011 21:36:28 EDT from the_mgt @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

Hehe, pretty ironic.

afaik, ClearOS decided to base their distribution (again) on RHEL sources directly in order to rely less on CentOS, not so much for full compatibility reasons. They were tired of waiting. Btw, as a google fanboy, you might like the idea of the appstore in the ClearOS webinterface. :-P

I will try and compare the 64bit versions of both distros in the next weeks, will give feedback then. (Also, how well the citadel rpms work!)



[#] Mon Jul 11 2011 13:02:02 EDT from LoanShark @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


cool. gonna start slurping centos 6 ASAP. RHEL is still the system I know, has long support lifecycles, is released when it's ready and not before, has good commercial support, and was built and hardened by the guys who pioneered the Linux implementations of several important enterprise technologies. That's enough for me.

[#] Tue Jul 12 2011 13:51:25 EDT from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

That's pretty much the set of reasons why we went with CentOS (with a potential upgrade path to RHEL if someone needs it) as our data center standard.

I have to admit that I am warming up to the Debian/Ubuntu way of doing things, but I would need a very compelling reason to have to get all of the support infrastructure in place, train the whole NOC staff, etc.

[#] Tue Jul 12 2011 15:40:24 EDT from LoanShark @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


/bin/service doesn't reproduce the boot-time environment, and should live in /sbin. Non-starter, I tells you!

[#] Tue Jul 12 2011 15:52:17 EDT from LoanShark @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


On the other hand, the customer is always right and if the customer wants Ubuntu... I dunno. But that might not be important; the broader question for you is, "how do I stack up against EC2 and RackSpace"?

With our EC2+RightScale stack, I can click a few buttons and spin up a new server, which is based on an AMI image plus a set of customizable installation scripts. I can click a radio button and choose whether those scripts run on top of CentOS or Ubuntu (various versions available) as the underlying image.



There are a few places in our environment where we're dealing with some distro-specific installation code in our scripts that we'd prefer not to have to deal with; one or two of our servers are running CentOS and the rest are on Ubuntu (wasn't my choice.)

[#] Wed Jul 13 2011 11:39:58 EDT from athos-mn @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

I've been given a problem by someone, and am curious how you all think this will work:

 

A client of the company I work for has an ancient Fedore Core 2 that we never backed-up properly (long story on that - thankfully it doesn't involve me) and has a custom program on it that's been retired, but is occasionally referenced. After two drives on the RAID 5 went south, the programmer, who was fired two years ago, told them to take a hike.

 

My plan is to take the drives to a recovery place (OnTrack) and see if they can deliver a coherent filesystem (as opposed to "here's a DVD with one segement of the array, here's a DVD with another - have fun putting them together!"), reinstall the OS (if nothing else to get the boot loader on it), then copy everything over - if not live than via a boot CD.

 

The last time I tried anything like this was in 1997 on a SCO OpenServer box, but I had a cpio backup - so everything else was in place for the restore.

 

Many thanks.



[#] Thu Jul 14 2011 10:25:14 EDT from Freakdog @ Dog Pound BBS II

Subject: Re:

[Reply] [ReplyQuoted] [Headers] [Print]

 

Wed Jul 13 2011 11:39:58 AM EDT from athos-mn @ Uncensored

I've been given a problem by someone, and am curious how you all think this will work:

 

A client of the company I work for has an ancient Fedore Core 2 that we never backed-up properly (long story on that - thankfully it doesn't involve me) and has a custom program on it that's been retired, but is occasionally referenced. After two drives on the RAID 5 went south, the programmer, who was fired two years ago, told them to take a hike.

 

My plan is to take the drives to a recovery place (OnTrack) and see if they can deliver a coherent filesystem (as opposed to "here's a DVD with one segement of the array, here's a DVD with another - have fun putting them together!"), reinstall the OS (if nothing else to get the boot loader on it), then copy everything over - if not live than via a boot CD.

 

The last time I tried anything like this was in 1997 on a SCO OpenServer box, but I had a cpio backup - so everything else was in place for the restore.

 

Many thanks.



I've had good luck with OnTrack, in the past...used it to recover a drive that had gone bad in an Netware 3.12 server.

While they're expensive as all get out, if the data recovery is really necessary, it's worth it.  When my personal server's drives bit the farm a couple of years ago, I took them to a local service, but they couldn't read anything that didn't come from a Windows system (last time I'll go with them...even if one of my family's Windows machines goes to hell).



[#] Thu Jul 14 2011 19:03:31 EDT from the_mgt @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

As an el cheapo alternative I can recommend Zero Assumption Recovery, it did really impressive recoveries for me, at least for various states of messed up NTFS partitions. They have linux fs support and even raid recovery and if you are lucky, you get your license for 50% on some weekends. (And I don't usually buy software (let alone use illegal copies) because there is often an opensource alternative.) But if money is  no problem and the drives are paleolithic, go for OnTrack or some other pros.

Link http://www.z-a-recovery.com/



[#] Fri Jul 15 2011 12:29:25 EDT from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

On the other hand, the customer is always right and if the customer
wants Ubuntu... I dunno. But that might not be important; the broader

question for you is, "how do I stack up against EC2 and RackSpace"?


The customer is not always right. The customer is in charge. That's not the same thing. It's our job to help them be right. At least that's the approach we take, and it is appreciated most of the time.

Stacking up against EC2 and Rackspace is a different issue altogether. We've never been able to compete with the retail grade web hosting places, and we don't intend to compete with the retail grade cloud providers either. It's not a game we can win.

[#] Sat Jul 16 2011 12:59:00 EDT from the_mgt @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

http://boxgrinder.org/ sounds really fancy. Anybody here tested it?

BoxGrinder creates appliances (also called images) from simple plain text Appliance Definition files. There are only two simple steps to create an appliance:
Create Appliance Definition File – Check out our appliance definition structure page for detailed information about the file structure along with examples.

name: jeos-postgresql
summary: JEOS based on Fedora 14 with PostgreSQL server
os:
name: fedora
version: 14
packages:
- postgresql-server

Run BoxGrinder – BoxGrinder will download all the necessary artifacts, build the instance, convert it to the selected platform and upload it to the selected destination. All in one process!

boxgrinder-build jeos-postgresql.appl -p vmware -d local

That’s all!

 



[#] Sun Jul 17 2011 10:22:34 EDT from fleeb @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


First I've heard of it. Sounds intriguing.

[#] Sun Jul 17 2011 13:38:10 EDT from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


The thing that I don't quite understand about appliance generators like boxgrinder, rpath etc etc is that they all seem to want you to specify packages and stuff and then they build your appliance.

Where is the part where you go in and tune things up, preconfigure the packages to work together, add all the customizations that are the reason appliances are so convenient in the first place?

If I were to build a Citadel appliance for example, I would want to not only install spamassassin and clamav but also preconfigure Citadel to use them.
The first thought would be "build that into the package installation scripts" but if the deb's are that smart, there's little point in building an appliance in the first place.

I want to build my appliance manually. Then I want to tune it so it's absolutely perfect. Then and only then, do I want it automagically converted so there are optimized downloads available for VMware, VirtualBox, KVM, etc etc.

I don't think anyone is doing that yet?

[#] Sun Jul 17 2011 14:56:41 EDT from the_mgt @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

Yes, that is where I am stuck, too. Maybe there is some sort of "patch set" applied to the image, where the default distro things are touched and the changes to the configs are made. If I need to chroot into the machine or even boot it and ssh into it, I could simply run self generated scripts or do stuff by hand when I am there...



[#] Sun Jul 17 2011 22:49:06 EDT from LoanShark @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

I want to build my appliance manually. Then I want to tune it so it's

absolutely perfect. Then and only then, do I want it automagically
converted so there are optimized downloads available for VMware,
VirtualBox, KVM, etc etc.

And AMI's for Amazon EC2. For that, you can do exactly what you're asking for via RightScale: launch a machine via one of RightScale's prebuilt images. Install packages. Customize. Click the "bundle image" button in RightScale, which snapshots the filesystem. Done.[1]

[1] I haven't actually tried this yet; one co-worker wants us to start deploying our machines based on custom images, for performance reasons; I prefer scripted installs, for repeatability reasons. Repeatability should always trump performace, until you just can't stand the slowness anymore.

[#] Mon Jul 18 2011 08:38:08 EDT from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

I agree about repeatability, and you're probably aware of my opinion there if you've seen Easy Install, which is not exactly a race car, but it tends to produce consistent results regardless of the underlying system state (or platform origin).

So I guess that means the goal is to write more scripts. It will be a much more tedious process to build the appliance but the results will be a set of scripts that can quickly be deployed to any appliance building host -- not to mention the additional result of "hey, here's a set of scripts you can download and run" :)

[#] Mon Jul 18 2011 09:23:34 EDT from IGnatius T Foobar @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]

You know, now that I think about this a little more, I already kind of went down that path. The appliance image I've already got out there (which is getting *very* long in the tooth at this point) contains a script that I wrote to prep the appliance for distribution. It zeroes out the swap space, deletes the logs, and does a few other things to clean up the mess left behind by me (the maintainer) logging into the appliance to perform updates.

So I guess it really does make sense to just move all this scripting stuff to the front of the line, and consider the "value add" of the appliance to simply be knowing what downloads to include and what order to run the integration scripts in.

Plus, of course, the "hey here are the scripts if you want to do it yourself" benefit for the ultra-crunchy geek set. :)

[#] Mon Jul 18 2011 10:35:30 EDT from LoanShark @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


If I were going to publish a Citadel image "on RightScale", without a requirement for building generic AMIs, I'd not even build an image, I'd just attach a set of boot RightScripts (which are just shell scripts with a specific input variable convention) to the server template, which would be responsible for installing Citadel. (RightScripts can have binary attachments.)

[#] Mon Jul 18 2011 11:07:02 EDT from LoanShark @ Uncensored

[Reply] [ReplyQuoted] [Headers] [Print]


If you live outside of the Amazon, you can also use tools like virt-rescue[1] to manipulate an appliance image without actually running it:

[1] http://libguestfs.org/virt-rescue.1.html

Go to page: First ... 61 62 63 64 [65] 66 67 68 69 ... Last