The Logical Volume Manager LVM2 in Linux is a great way to become independent from physical – or virtual – disk devices.
Especially in a cloud environment like the Amazon Web Services where you can add or remove virtual disks (called “Elastic Block Storage”, EBS) at any time, LVM makes it easy to adapt a server’s storage capacity to current demand.
Preparations: Create an LVM volume
To build your filesystem, you first create an EBS volume via the Web Console or the EC2 command line tools and attach it to the EC2 instance. The default block device name is /dev/sdf, but you can choose another one if you want or need to.
This EBS volume must be marked as a Physical Volume (PV), the basic building block of our LVM storage pool:
# pvcreate /dev/sdf
The PV becomes the foundation of a Volume Group (VG). Later, you’ll be able to add other PVs to the VG, thereby growing the VG’s storage capacity.
# vgcreate vg0 /dev/sdf
Finally, from this VG, you create Logical Volumes (LV). LVs are the virtual disk devices that you’re going to format and mount. Modern filesystems like ext3/4 or XFS are able to grow in size without reformatting; this will come in handy later.
# lvcreate -L20G -n vol1 vg0 # mkfs.xfs /dev/vg0/vol1 # mkdir /vol1 # mount -t xfs /dev/vg0/vol1 /vol1
How to increase a volume’s size
If your virtual disk, the LV, is filling up, you can easily grow its capacity. Just create and attach another EBS volume (for example as /dev/sdg), add it to the VG, and finally extend first the LV and then the filesystem:
# pvcreate /dev/sdg # vgextend vg0 /dev/sdg # lvextend -L+20G /dev/vg0/vol1 # xfs_growfs /dev/vg0/vol1
How to replace an EBS volume
Alternatively, you could just replace the current EBS volume with a bigger one. This is easy because LVM lets you add a bigger EBS volume to the VG, then migrate all data onto it and finally remove the old volume from the VG:
# pvcreate /dev/sdg # vgextend vg0 /dev/sdg # pvmove /dev/sdf /dev/sdg # lvreduce vg0 /dev/sdf # lvextend -L+20G /dev/vg0/vol1 # xfs_growfs /dev/vg0/vol1
Now, you can detach the old EBS volume from the EC2 instance and then delete it.
We use virtual IT infrastructure resources from the cloud for their flexibility. By using LVM, we maintain this flexibility even on the single filesystem level.
17 Nov 2011
Last week was a special one.
Jochen had the first week off after a very long time without vacation. And I had to take care of all the work alone. I knew it would be a lot of work, but I underestimated almost everything.
As a result, the only day on which I could do “normal work” like site building for a customer was Thursday. On the days before, I was overwhelmed by support requests, coming in either via email or phone calls. I really wanted to do this job as best I could, so I guess I forgot my organizational skills and put support requests above everything else. Which resulted in some happy customers, but an exhausted me.
Since this was a really demanding week, we’ll have to learn from that and organize some stuff differently. Customer support has to work better even if one central person is missing, and all other, no less important, work has to keep going on as well. We will implement some rules and processes to improve our workflows and make sure that both our customers and ourselves stay happy.
In the end it actually was rewarding as well, having gotten so much positive feedback. Thanks for that!
15 Nov 2011
In our weeknotes, we’d like to give you a short recap into what’s been happening at freistil during the week. The numbering is based on the age of the business in weeks and because the latest incarnation of freistil IT officially started on April 6th 2010, this is Weeknote #83.
It’s been a long time since my last weeknote. Over the recent months, we’ve been very busy and traveled a lot. We took part in three Drupal conferences , namely DrupalCon Europe in London, Drupal Commerce Camp in Lucerne and DrupalCity in Berlin. Each and every time, we had the pleasure of meeting customers we hadn’t met in person yet as well as making new aquaintances. Having many opportunities to talk to people proved at least as valuable as the conference sessions where people talked to us. It’s really fun to be part of such an amazing community.
A huge part of our work is done behind the curtains. For example, we continuously improve our infrastructure management . We monitor an increasing number of servers (currently more than 70 dedicated machines) and services (more than 1000 individual checks every few seconds). Switching from a run-of-the-mill SMS service to PagerDuty allows us to effectively alert the on-call engineer and escalate an incident to the second level, if necessary. At the heart of our infrastructure management, we use Chef to automate all our system administration tasks. There is no day we don’t add new “recipes” or improve existing ones. Jochen recently built a dedicated storage cluster for our central backup system. Based on a distributed GlusterFS filesystem, this storage cluster can be expanded by simply adding new storage servers while a rebalancing process makes sure that files are spread evenly on all storage nodes.
Of course, we’re also heavily improving our products. Especially our DrupalCONCEPT ELITE customers that pay for an all-inclusive, highly scalable hosting infrastructure benefit a lot from many performance and stability improvements .
Because we are very aware that our daily workload must not prevent us from planning ahead, we’re doing quarterly two-day strategy meetings . Our last get-together was at the beginning of October and we made some important decisions for the coming year. One of those was to rebrand the company under the name “freistil IT” (see our announcement).
We also defined a new hosting product line that will close the price gap between our entry-level products and our offerings for higher demands. These new hosting products will be based on our own private cloud , i.e. virtual servers we can quickly provision with exactly the system resources the customers need. By operating this private cloud ourselves, we benefit from higher flexibility as our customers do from lower prices. (Infrastructure projects like this make sure that Jochen doesn’t get bored with just rebooting servers from time to time.)
A feature our hosting customers have been requesting for a long time is a self-service dashboard . After deciding to ditch Ruby on Rails and build upon Drupal (as was the original idea in the first place), Markus is now making rapid progress and we’re aiming for a first production version before Christmas. This is a very important project for us because the more settings our customers can configure themselves without our help, the more time we gain to work on value-adding projects.
We’ll try to post weeknotes now on a more, well, weekly basis. To give you an interesting insight, we’ll be taking turns (tip o’ the hat to the guys at Third Wave). I’m already curious what Markus will be telling next time!
04 Nov 2011
Many customers prefer to test changes to their websites before they deploy them to the production website, so they ask us how they can best implement a staging workflow on our DrupalCONCEPT Drupal hosting platforms.
We recommend to have a three-stage workflow, as we’ve explained in our session at the DrupalCity conference in Berlin:
- Development environment
- Pre-Live test environment (often called “staging environment”)
- Production environment
In the following Help Center articles, we’ve explained in detail how to implement this three-stage workflow:
- Sicheres Ausrollen von Änderungen mit Hilfe von Staging-Umgebungen
- How to use staging environments to securely roll out changes
27 Oct 2011
When we started our DrupalCONCEPT hosting products in 2010, incoming phone calls increased by an order of magnitude due to our support hotline. We would start every call with something like “Freistil-Consulting, my name is Jochen Lillich. How may I help you?” and ask many callers to send us their support request to “support at freistil dash consulting dot de”. Over time, we noticed that our company name and domain could be simpler and we started searching.
Because “freistil”, the german word for “freestyle”, greatly fits our modus operandi, we decided to keep this part and just drop the “consulting”. And since IT is what we’re good at and doing day to day, we went for “freistil IT” and the nicely short domain “freistil.it”.
Lean operations is the foundation of our company and so we decided to make the new website that had to come with our shorter name as lean as possible as well. Nothing more than a home page, a contact page, some information about our team and a blog. Everything in one language understood by our customers all over Europe.
From now on, the blog will be our main information channel that then feeds into other channels like Twitter or our newsletter. On the blog, we’ll regularly post company news, product innovations and tech articles. So, to stay up-to-date on all things “freistil”, you can simply subscribe to our blog’s RSS feed, to our email newsletter or follow our company Twitter account @freistil.
Also posted on the blog will be our “weeknotes”, weekly(ish) recaps of what happened at freistil IT. Jochen used to post them on his personal blog, but since they’re about the company, we think they belong on our company blog.
We’ve been working hard behind the scenes over the recent weeks. So hard, actually, that we didn’t have enough time left to tell you about the new features of our DrupalCONCEPT hosting products we’ve been working on. We’ll catch up as quickly as possible.
And it’s not only details we have improved: There will be a whole range of new DrupalCONCEPT products that we’ll unveil in a few days!
So, keep watching exactly this very spot for exciting product news and interesting tech information — brought to you by freistil IT!
21 Oct 2011