Flickr added a new thing where you can print five pictures and fold it into a cube. Much fun!!
Friday, November 30, 2007
Wednesday, November 28, 2007
Day 184: IM in UR candy eating UR... uhh candy
The statute of limitations has run out on Halloween candy. At this point, if I find it, it's mine. It's good to be the king.
Heartbeat 2 test platform guide closer
I finally got to the point where heartbeat gets installed. I'm learning a bit on the way. For example, you want to install heartbeat -after- you clone your virtual machines due to heartbeat generating it's own unique ID for the node in the cluster. The documentation is done up to the point of installing heartbeat 2 on a three node cluster. If you get to the end and run 'crm_mon -i5' you should see something like this:
============Now I need to create resources, and test some basic failovers.
Last updated: Wed Nov 28 16:44:29 2007
Current DC: deb1 (b4e5f810-d4fc-449f-bab3-c2b0aefbfa1a)
3 Nodes configured.
0 Resources configured.
============
Node: deb1 (b4e5f810-d4fc-449f-bab3-c2b0aefbfa1a): online
Node: deb2 (fe44e854-cb58-4087-9317-e12ae227fc50): online
Node: deb3 (6d5ec010-1c89-4876-bd9c-b92dca1b51e8): online
Tuesday, November 27, 2007
Heartbeat 2 testing grounds
I'm currently working on a "cut-and-paste" guide for building a playground for Heartbeat version 2. This guide starts with install Vmware Server, and building a minimal debian virtual machine. The I go into making an iSCSI target and initiator. Then we format that with OCFS2, and then install heartbeat-2.
That's as far as I got. I'm working on cloning the debian virtual machine into two more nodes giving us three nodes to play with. I linked the guide to the right of this page.
Linux-HA full testing guide - Test out Heartbeat 2
That's as far as I got. I'm working on cloning the debian virtual machine into two more nodes giving us three nodes to play with. I linked the guide to the right of this page.
Linux-HA full testing guide - Test out Heartbeat 2
Day 183: Making rock candy
My son is trying this experiment for the second time. We're trying to make rock candy. We think we failed the last time because we did have seed crystals on the string before. I'm not sure it's going to work this time as I think the water needed to be colder. I can't see the seed crystals either so I think they dissolved. We might not be getting enough sugar into this solution.
Monday, November 26, 2007
Day 182: Playing with VMWare
I'm building a how-to for making a virtual cluster with linux HA. I think I have the iSCSI targets all set and I format them with OCFS2. Now I just need to get the Apache service to fail over to other nodes and document it.
Sunday, November 25, 2007
Saturday, November 24, 2007
Day 179: My black Friday prize
I try to just get the loss leaders and stay away from everything else. 2gig SD for $13.
Wednesday, November 21, 2007
The broken troops
I was reading a blog post about wounded soldiers returning home to a hero's welcome at the Pentegon. (Friday Mornings at the Pentagon)
It's great to read that at least some people at the top clap, cheer, and cry for these people. The same folks are also going into debt to repay the sign on bonus they received. Because they got blown up and lost that leg, they are unable to serve out the rest of their enlistment. Therefore, they have to give back the money the military gave them to sign up. In some cases $30,000.
Just one story here: Wounded Vet Told To Pay Back Bonus
So welcome home! Thank you so much for your service! I shed tears to see you with your legs blown off in that wheelchair, but uh... can I have that money you owe me? You didn't actually fulfill the contract, so... when can I expect my money?
It's great to read that at least some people at the top clap, cheer, and cry for these people. The same folks are also going into debt to repay the sign on bonus they received. Because they got blown up and lost that leg, they are unable to serve out the rest of their enlistment. Therefore, they have to give back the money the military gave them to sign up. In some cases $30,000.
Just one story here: Wounded Vet Told To Pay Back Bonus
So welcome home! Thank you so much for your service! I shed tears to see you with your legs blown off in that wheelchair, but uh... can I have that money you owe me? You didn't actually fulfill the contract, so... when can I expect my money?
Tuesday, November 20, 2007
Day 177: My first snowball of the season
Yea, I'm at work. I wanted to get a photo of it before all the snow melted.
Sunday, November 18, 2007
First look at Vmware Server 2 (beta)
It sucks! Well, I'm probably just using it wrong, but can't for the life of me find the way to get into the flipp'n BIOS. I did it one time, and I'm not sure how that happened. Might have been the first time the VM was booted. I was reading that the default behavior is for the VM to boot off a CDROM the -first- time it boots up. After that first boot, it switches to booting from the hard drive. If you messed up and need to boot the CDROM again, you seem to be hosed. So I had to build new machines... and then found out that I can't seem to be able to delete the old junk VMs. They just sit around in the list forever. Even if you go to the directory and delete it, the remains hang around like a ghost, forever populating your inventory. I'm sure there is some whacked out config file that I can hex edit to remove these things, but I want to do it through the interface. After all, I created it there really easily.
Vmware Server 2, with it's web interface really acts a lot like the Virtual Center software that I've been using at work. It's not all that bad, though. It does seem like a better server than the older version, but I think the web only interface make it harder to get used to. I am still right clicking on item is hopes of finding a properties menu or something.
Vmware Server 2, with it's web interface really acts a lot like the Virtual Center software that I've been using at work. It's not all that bad, though. It does seem like a better server than the older version, but I think the web only interface make it harder to get used to. I am still right clicking on item is hopes of finding a properties menu or something.
Day 175: VMWare is pissing me off.
I'm trying out the new VMWare Server 2 which is all web based now. I'm having one hell of a time with getting into the BIOS on this thing, and I can't seem to be able to remove dead systems from the list.
Friday, November 16, 2007
Thursday, November 15, 2007
Day 172: Mini Walmart
I saw this vending machine in the Dallas airport. Neat! It's like a mini Walmart
Here is the full view:
Here is the full view:
Wednesday, November 14, 2007
Next generation configuration management?
I attended a talk (BOF) last night, while here at LISA, about configuration management. One struggle is that the name gets confused with change management. What we are talking about is using tools to configure and then maintain that configuration across many systems. So you can just create a web server configuration, and from that you can configure all your web servers automatically. This make some admins from small shops nervous, but if you are a hosting company like Rackspace, or a company like Google, all the automation possible creates more uniformity and less chance for human error. It also frees up the admin to perform actual trouble shooting instead of the robotic work of building servers.
The tools discussed last night were CFEngine, Puppet, and Bcfg2. One of the big struggles is getting people to understand and to use the tools. I must admit, I struggle with using CFEngine. I don't want to go and check in configs into subversion and all that. It seems like more work than just going to the system and fixing or changing the little thing I need to change. The issue then becomes that I'll keep doing that kind of change as long as it doesn't exist in CFEngine. One person used the analogy of compiling a build. The configuration tools are like the compiler. We edit the source code (central config files), and then build the binary (push the configs). We don't go in and try to edit the binary file after it's been built. You go back into the source code, make your changes, and build it again. The same method has to be taken with configuration management. Get into the habit of -not- configuring your leaf nodes, but pushing the config to them.
One of the more interesting things, in my opinion, that was thrown around was the talk about where configuration management was going next. After some talk, someone brought up that he'd like something similar to what LDAP did for logins/passwords. This really stuck in my head. Using something like LDAP for system configurations. I really hadn't thought about using LDAP for much beyond account information. I'm still sitting back contemplating the extraordinary thing I might be able to do with something like LDAP.
The tools discussed last night were CFEngine, Puppet, and Bcfg2. One of the big struggles is getting people to understand and to use the tools. I must admit, I struggle with using CFEngine. I don't want to go and check in configs into subversion and all that. It seems like more work than just going to the system and fixing or changing the little thing I need to change. The issue then becomes that I'll keep doing that kind of change as long as it doesn't exist in CFEngine. One person used the analogy of compiling a build. The configuration tools are like the compiler. We edit the source code (central config files), and then build the binary (push the configs). We don't go in and try to edit the binary file after it's been built. You go back into the source code, make your changes, and build it again. The same method has to be taken with configuration management. Get into the habit of -not- configuring your leaf nodes, but pushing the config to them.
One of the more interesting things, in my opinion, that was thrown around was the talk about where configuration management was going next. After some talk, someone brought up that he'd like something similar to what LDAP did for logins/passwords. This really stuck in my head. Using something like LDAP for system configurations. I really hadn't thought about using LDAP for much beyond account information. I'm still sitting back contemplating the extraordinary thing I might be able to do with something like LDAP.
Tuesday, November 13, 2007
Day 170: Free Beer and Ice Cream
There was an event that gave away free beer and ice cream. Yummy! I don't think anyone really paid attention to the struggling sales person trying to do a slide show.
Monday, November 12, 2007
My thoughts on Cfengine at LISA 07
Lecturer - Mark Burgees - Oslo University College
Lecture 1: Integrating Cfengine info Organizational Service Management
Lecture 2: Implementing Some Autonomic Principles Using Cfengine
One of the biggest things I have noticed about Cfengine is that it has close ties to ITIL (Information Technology Infrastructure Library). There was also a lot of discussion for other compliance policies like Sarbanes-Oxley, but the main focus by the author was ITIL. I think that getting familiar with ITIL will lend some understanding to what some of the things Cfengine is trying to accomplish.
One thing to note about the current state of Cfengine is the migration to something called "promises". Version 3 promises (no pun intended) to use these in it's language much more. That is to say, instead of telling the server what to do, the server makes promises to what it's going to do. It's more a shift in thinking about how Cfengine is perceived.
Something that I didn't realize was by making changes in Cfengine, you get to have a history of changes also known as an audit trail. This is accomplished by taking advantage of the versioning built into Cfengine, at least in v2.2.2.
The other big thing was the relation to political philosophies pertaining to the management style. There was the start, distributed star, and then peer-to-peer methods mentioned. Based on how we run things back in the shop, I figured the distributed star was the best, while the star gave a single point of failuer, and the peer-to-peer was simply chaotic. First, while we all understand that there are internal variables to the systems like disks, RAM, and processors. We might even take into account the network bandwidth or even network location, we'll generally not take into account other factors like the environment. One big environment variables is the user. The author suggests that no system knows the capabilities of the server, than the server itself. Let the server decide for itself the work load, and it's config. This autonomous style is obviously much closer to the Libertarian principles, while the central star topology is more communist. The peer-to-peer style becomes much more obvious as you apply autonomic principles to using Cfengine.
Another important nuance the author pointed out was idea of "rollback". He's not to fond that that kind of thinking. The reason is because even though you can go back to a previous config, there are still environment variables that don't roll back. Because of this, it's faulty thinking to consider a rollback instead of simply fixing what ever is wrong and keep moving. The analogy used was that of a ship at sea. To roll back, you have to stop, go in reverse, and get back to the rollback point. But even though you go back, you might drift in the current making it very hard to get there. Instead of going in reverse, keep moving forward, but turn the boat around. This way you are always moving in a forward direction.
Lastly, some things I took away from this session was that there are some tools in Cfengine, like cftool that can provide graph data. I need to study this more, and take advantage of this. The other is letting Cfengine handle more of the crontab files. This might solve some of the sync issues that have come up in some of the cluster systems.
Related Links:
http://www.cfengine.org/
http://en.wikipedia.org/wiki/ITIL
http://en.wikipedia.org/wiki/Sarbanes-Oxley_Act
Lecture 1: Integrating Cfengine info Organizational Service Management
Lecture 2: Implementing Some Autonomic Principles Using Cfengine
One of the biggest things I have noticed about Cfengine is that it has close ties to ITIL (Information Technology Infrastructure Library). There was also a lot of discussion for other compliance policies like Sarbanes-Oxley, but the main focus by the author was ITIL. I think that getting familiar with ITIL will lend some understanding to what some of the things Cfengine is trying to accomplish.
One thing to note about the current state of Cfengine is the migration to something called "promises". Version 3 promises (no pun intended) to use these in it's language much more. That is to say, instead of telling the server what to do, the server makes promises to what it's going to do. It's more a shift in thinking about how Cfengine is perceived.
Something that I didn't realize was by making changes in Cfengine, you get to have a history of changes also known as an audit trail. This is accomplished by taking advantage of the versioning built into Cfengine, at least in v2.2.2.
The other big thing was the relation to political philosophies pertaining to the management style. There was the start, distributed star, and then peer-to-peer methods mentioned. Based on how we run things back in the shop, I figured the distributed star was the best, while the star gave a single point of failuer, and the peer-to-peer was simply chaotic. First, while we all understand that there are internal variables to the systems like disks, RAM, and processors. We might even take into account the network bandwidth or even network location, we'll generally not take into account other factors like the environment. One big environment variables is the user. The author suggests that no system knows the capabilities of the server, than the server itself. Let the server decide for itself the work load, and it's config. This autonomous style is obviously much closer to the Libertarian principles, while the central star topology is more communist. The peer-to-peer style becomes much more obvious as you apply autonomic principles to using Cfengine.
Another important nuance the author pointed out was idea of "rollback". He's not to fond that that kind of thinking. The reason is because even though you can go back to a previous config, there are still environment variables that don't roll back. Because of this, it's faulty thinking to consider a rollback instead of simply fixing what ever is wrong and keep moving. The analogy used was that of a ship at sea. To roll back, you have to stop, go in reverse, and get back to the rollback point. But even though you go back, you might drift in the current making it very hard to get there. Instead of going in reverse, keep moving forward, but turn the boat around. This way you are always moving in a forward direction.
Lastly, some things I took away from this session was that there are some tools in Cfengine, like cftool that can provide graph data. I need to study this more, and take advantage of this. The other is letting Cfengine handle more of the crontab files. This might solve some of the sync issues that have come up in some of the cluster systems.
Related Links:
http://www.cfengine.org/
http://en.wikipedia.org/wiki/ITIL
http://en.wikipedia.org/wiki/Sarbanes-Oxley_Act
Day 169: OMFG It's Tobias!!
This is the Tobias Oetiker, creator of MRTG and RRDTool! Not many people grasp how huge of a deal this is. Both of these simple graphing tools are at the roots of so many other open source projects. This guy right here is responsible for so much.
And personally, it was these graphs that I used to show the boss a graphical and simplified view of what was going on with the computers. The management liked it so much that it became part of what I do from then on. Taking systems metrics and making pictures for managers. This guy next to me keeps me employed.
So yes, I was very excited to meet him, and tried very hard not to shake the camera, but I was a bit nervous. I even told him I was a huge fan. Something tells me he's not used to that kind of stuff.
On a different note, during my Solaris class I somehow managed to take down the Usenix wireless network... because I was downloading the opensolaris ISO images as well as VMWare Server. Ooops. =)
And personally, it was these graphs that I used to show the boss a graphical and simplified view of what was going on with the computers. The management liked it so much that it became part of what I do from then on. Taking systems metrics and making pictures for managers. This guy next to me keeps me employed.
So yes, I was very excited to meet him, and tried very hard not to shake the camera, but I was a bit nervous. I even told him I was a huge fan. Something tells me he's not used to that kind of stuff.
On a different note, during my Solaris class I somehow managed to take down the Usenix wireless network... because I was downloading the opensolaris ISO images as well as VMWare Server. Ooops. =)
Sunday, November 11, 2007
Day 168: Me and Mark Burgess
This is -the- man who created CFEngine. There were some fascinating talks. I still feel very strange asking people to take a photo with me. I'll have to get over it as I plan on taking more. I gotta remember to smile.
Saturday, November 10, 2007
I arrived at the conference
Only to find out that my company didn't actually pay for the classes. Oh they signed the permission, and authorization, and purchase order, and got it cleared through the twelve layers of bureaucracy... But no one actually sent in the check!
Day 167: On the plane
The kid right next to me was William and the kid right behind him (his younger brother, Parker) pretty much annoyed each other the whole trip. William almost received a pummelling (from me) when he started yelling at Parker, telling him to stop kicking his seat. His grandma in the aisle row got him before I could take action.
Friday, November 9, 2007
Day 166: Packed for the nerd conference
I'm so excited. I use and abuse a lot of open source software in my job and some of the people that created these tools are going to be speaking at the conference I'm going to, like Tobias Oetiker (creator of RRDTool). I'm going to try and get a 365 with some of these people. Google is going to be giving a talk about infrastucture and I'm just totally geeked to hear it.
I hope I packed enough nerd supplies.
I hope I packed enough nerd supplies.
Monday, November 5, 2007
Day 162: Mmmm, Garlic Bread
Ihad stuff to do so I didn't get home in time to get the fresh warm bread (which is why the butter isn't melted in this picture), but even so, it's still yummy.
Sunday, November 4, 2007
Saturday, November 3, 2007
Friday, November 2, 2007
Day 159: Googly Eyes
My boss at work got a bunch of these gummy eyes. I thought I'd come home and scare the kid with them.
Subscribe to:
Posts (Atom)
Posts
