Tag Archives: media temple

Hosting: where to go, where to go

18 Aug

So after about 2 months on slicehost, I have yet to finish setting up my server. It seems to be the busy season and I’ve only been able to finish my local servers.

So to recap I know have a media temple box and a slicehost box. Now to further complicate the situation after much debate / research I’m looking at linode: http://www.linode.com.

Why linode?

  1. They use Intel processors
  2. More ram, space, bandwidth
  3. better control panel
  4. better vps management

That said slicehost has been great for information. I’m not a fan of their forums but they do have great tutorials.

Why slicehost?

  1. They seem slightly more stable
  2. Made by developers for developers
  3. First impressions

…. so that said, sorry for the lack of new tutorials, code, broken templates, etc

And now to see about canceling one of these damn server

Slicehost

10 Jun

I’m been an MediaTemple fan boy for awhile, but I’ve been looking for a new host for the past few months. Today I may have found them: slicehost.

Their vps servers are called slices, and that what you get. Each slice is a base install of your os of choice – nothing else. You setup the services / apps you need, nothing more, nothing less. Honestly what got me was their community. Not only are staff always somewhere to reach them, but the users are seriously active and helpful. Oh yeah and their tuts rock!

So yeah whenever I get the chance all my unused domains, friends domains, etc will be moving to my the new slice…. Mt is still good, but i’d rather have a bit more control, and pay less than half the price for it

Install yum and subversion on media temple dv 3.5

29 Mar

Just upgraded one of my clients servers to dv 3.5. dv 3.5 uses centos 5 so when you install yum you’ll need to use that version

 

rpm -Uvh http://mirror.centos.org/centos/5/os/i386/CentOS/yum-3.0.5-1.el5.centos.5.noarch.rpm

Subversion is now included in the developer tools package. Login to your account center and install developer tools from there.

via hello, hellm 

deploying rails made simple : Deploying Rails From Scratch

19 Mar

I spent a good bit of this summer working on websites for ApacheCon US 2007 and OS Summit Asia 2007 which will be in Atlanta and Hong Kong this fall. (You should attend!) There’s a whole long sad story behind the Apache conference websites, but I’ll spare you that (for now). Instead, I wanted to collect in one place all the steps and articles you need to get Rails up and running on a Media Temple dedicated virtual server.
Our Goal

Our goal is to have a Ruby on Rails applicatin running on a Mongrel cluster proxied by Apache httpd. We’ll use Capsitrano to automate the deployment process. In the end, we’ll be able to issue a single command to tag the release in subversion, check it out on the server, and deploy the new code. We’ll start with a bare dedicated virtual server.

Deploying Rails From Scratch :: Cubicle Muses

media temple nitro

13 Mar

I finally got my hands on a nitro, old news, but super sweet, had to let the world know

How to: Install electroserver on media-temple dv and similar centos setups

11 Mar

Before you start

Before starting anything you should be sure that you have the developer tools package installed. If your not sure, then you don’t. No worries: submit a request to install the developer tools package and you’ll be ready in about 45min to a hour.

Get started

first login as root

next download electroserver:

wget http://www.electro-server.com/downloads/builds/ElectroServer_4_0_4_linux.rpm

next install the rpm:

rpm -iv ElectroServer_4_0_4_linux.rpm

You should now have a /opt/ElectroServer_4_0_4/ directory. browse around it and we’ll continue from there.

I had no issues up to this point, but feel free to post any problems and I’ll try to help.

Testing the install

in the /opt/ElectroServer_4_0_4/ you’ll notice a few files. to test your install run the ElectroServer command

If your lucky you’ll have no errors or issues getting the server running. I encountered an issue related to memory available. a quick reboot of the server fixed this.

Configuring the server

For this tutorial we’ll use port 8080 for the web admin, and port 9898 for our chat server. if you need to use other ports just follow along. you can also repeat this process to enable additional ports.

we’ll need to enable access to port 8080 and 9898. port 8080 will be used for the web admin, port 9898 will be used for the text (chat) connection.

to open your ports run the following:

port 8080

/sbin/iptables -A INPUT -p tcp --dport 8080 -j ACCEPT
/sbin/iptables -A OUTPUT -p tcp --dport 8080 -j ACCEPT

port 9898

/sbin/iptables -A INPUT -p tcp --dport 9898 -j ACCEPT
/sbin/iptables -A OUTPUT -p tcp --dport 9898 -j ACCEPT 

next edit your ES4Configuration.xml which can be found at: <install directory>/server/config. if your following this tutorial on a media temple dv server this directory is /opt/ElectroServer_4_0_4/server/config/ES4Configuration.xml

ES4Configuration.xml your default configuration should look like this:

<?xml version=”1.0″ encoding=”UTF-8″ standalone=”no”?><ElectroServerConfiguration>

<!– This is the name of the node –>
<Name>StandAlone</Name>

<!– Defines the host and port on which the web server will listen for inbound connections –>
<WebListeners>
<WebListener>
<Host>127.0.0.1</Host>
<Port>8080</Port>
<Ssl enabled=”true”/>
</WebListener>
</WebListeners>

</ElectroServerConfiguration>

to get our server live we’ll need to give it an ip address, change 127.0.0.1 to an ip that is live on your server

Getting the server running and listening

before we can actually test our install we need to make sure that we’ve placed a crossdomain.xml at the root of the website listening at the ip used. if the ip is shared place the crossdomain file in the root of the default domain. the crossdomain.xml should be viewable at http://<electroserver ip>/crossdomain.xml, replace <electroserver ip> with the ip used in your ES4Configuration.xml.

be sure that you create allows for your running host, and any other domains that will access the server.

now restart electroserver if its already running and test if electroserver is excepting connections using telnet.

test that electroserver is running and able to listen by running telnet <electroserver ip> 8080

telnet should successfully connect. if not try shutting down all electroserver processes and start electroserver again. we’re now ready to get moving with getting our chat server running.

Use electroserver admin

the configuration used in this guide uses ssl, if your not using it then remember to use http where the guide says https.

access admin at https://<electroserver ip>:8080/admin

default user name: administrator
default password: password

once logged in go to gateways and edit the default gateway. change 127.0.0.1 to <electroserver ip>, save and restart the gateway

Test Chat Access

Electroserver by default runs chat on port 9898. restart electroserver and telnet to your electroserver ip, this time testing port 9898. your should recieve the contents of your crossdomain file as a response.

Almost Done

we’ll also need to add electroserver to our server startup. First create a file named: StartElectroServer, next  add the following to that file:

/usr/bin/nohup /usr/local/bin/ElectroServer &
exit

Place the script in you opt/electroserver directory with read and execute permissions. Now the final step link your file so it’s called at startup

ln -s /opt/ElectroServer_4_0_4/StartElectroServer /etc/rc.d/init.d/StartElectroServer

thats it, you should now be all set to run electro server.

Also to run electroserver without it closing when your console closes use: nohup ElectroServer &

Got questions? post a comment, or just post a comment to let me know this was helpful.

Sid

from the trench: streaming a live event

6 Nov

from the trench: streaming a live event

I decided to document my last event. don’t know what this falls under, but hopefully it helps others.

so lets get started

the event

I was contacted by the client to do a live webcast for a horse show. I would setup everything needed prior to the event, and also come to the event location for a day and a half to insure everything ran smoothly, was setup, etc.

the setup

We used fms installed on a media temple extreme server. The extreme server was to be used initially because we didn’t know how many users would be watching. if and when traffic to the feed started to elevate we’d contact mt to upgrade to a nitro server.

for broadcasting we used flash media encoder

day one

I arrived at the event to learn we still needed to pick up some of the equipment to access the video feed. We got to circuit city (no fry’s nearby) and decided on the WinTV-HVR-950 model 1145 from hauppauge.

after arriving at the event, this is the day before the actual event, we found out the a/v people wouldn’t arrive till the morning of the event :( so the day was pretty much spent making sure everything was running, installed, etc.

day two – seehorse politics

day two started off badly to say the least. the company doing the video “seehorse” apparently had some issues because they had originally wanted to do the web feed. this lead to us receiving the feed used for internal broadcasting which completely sucked and had no audio, we dealt with this the best we could.

luckily they also had brought in outside video people to assist with the filming of the event. Namely Thomas “Rick” Rommarito. I can’t say enough about Rick Rommarito. Rick saw our feed and told us not to worry. Rick ran us a direct feed to the video, audio, and was a great help. keep in mind that seehorse video’s staff had pretty much told me and the company that had hired me to blow off (with out saying it of course). we we’re pretty much ignored by the seehorse video team. so thank you again to Thomas “Rick” Rommarito for everything he did. He is the only reason we got a good video feed or audio. If you found this article while searching on seehorse, please don’t use them, contact rick directly at : rickbommarito[at][sbcglobal][dot][net]. Very knowledgeable a/v guy.

day two – the server

so half way through the day we finally had a good video and audio feed, after a few compatibility issues and tweaking we we’re up and running. here are screen shots from both fms admin, and top to show you how we’re on resources. the cpu usage for the fms process is wrong, so look at the actual cpu usages. for any one interested we maintained about 220-230 concurrent users throughout the day. once that number started to increase is when the server started having issues (end of day). I’d say keep users at about 200 concurrent if your going to be on an extreme.

early in the day we realized we’d need to upgrade to the nitro asap. I contacted mt, and was told we could be upgraded by the following morning. near the end of the day when i contacted mt again for an eta we found out all nitros were sold out for at least a week.

*update* – turns out there was a communication issue. one of the techs had a nitro on standby, but the night shift wasn’t in the loop. found this out the following week

edgecast to the rescue

with our upgrade plan now shot we needed a new host. after limelight told us there’s a 7 day wait, and Akamai seems to have outsourced their entire operation. speaking with someone at Akamai was just pointless, not to mention I didn’t hear back from a sales rep for two days, and that was also outsourced.

*update – I received a call from their guys two weeks later, asking what type of service i want!*

Edgecast was a different story, they knew we we’re 3 steps from being screwed and had us up and running within hours, I’ve already spoken on them in a previous post, but i wanted to make sure i mentioned them again. it normally takes a few days to get everything setup, but james seriously helped us out.

day three

day three had us with a brand new video feed, a new feed to computer setup, and edgecast streaming.

first off the edgecast streaming is great, any future events will use them, screw maintaining our own fms servers.

by the end of the day everything was finally running smoothly, and i got a flight heading back home

issues
our biggest issue was video, which luckily for us Thomas Rick Rommarito handled that.

next up was our server, we’d previously spoken to mt to make sure the nitro was ready on standby. *update – see previous update * I can’t be mad at mt though, as they have proven over and over again to be rock solid and caring. they even contacted me twice after about the nitro issue. edgecast was amazing. their support, network, and customer service was amazing. both myself and the client we’re happy with them, and now the client will be rolling out an entire service using the edgecast system, which leaves me happy to build with their system again.

after those two major issues the reoccurring issue was stream quality. horse shows are very motion intensive, and although we had a good feed, we didn’t have good lighting at all. i recommend any hauppauge product, because I use them at home, and every other encoder tested lacked the features it had. we tryed a few different one’s at the clients request, but in the end always went back to the hauppauge.

to overcome some of the lighting issues i used hauppauge’s built in features to adjust the feed which helped produce a better feed. the client had purchased a number of other encoders because he wanted to split the video / audio with another device. I’ll never again use a pyro product. once we started using the pyro adapter it created nothing but issues.

so once we had the hauppauge device back as the video source i was able to tweak settings and get a decent quality stream that had little lag, good motion, and decent to to good quality. I found the sweet spot to be at a 380 bit rate. the client wanted clear motion, but also needed the horses and people to be of good quality also staying around 380 offered the best motion and quality without being to bandwidth heavy. we broadcasted at a custom 16:9 resolution.

another issue was that the client had an outsource team that developed, and maintained the broadcast site. any changes had to go directly through them. they’re a good group, but often the need to contact them for changes, as well as time differences we’re frustrating. I tried to use the player the client already had in place, but in the end just hacked out a quick and simple player when i finally got completely frustrated.

in the end

so what did i learn? motion intensive events are a different experience than non motion intensive ones. don’t use pyro products. use edgecast, the quality of our stream when using their network was worth it.

I’ve normally ran our own fms server for broadcasts events. I won’t ever again, the cost and benefits of edgecasts service just makes setting up and maintaining a fms server pointless. It’s less of a headache, and you don’t need to monitor anything. In the end the cost of using edgecast is less when you add the man hours, maintenance, headaches, chain smoking, etc.

media temple dv extremes are good for around 200-210 concurrent users, after than your gonna start feeling it.

also something I realized : widescreen formats (16:9) are better then 4:3 for live streams. I found that when using a wide screen format the viewing experience was alot better, especially at smaller sizes than traditional 4:3 format. i was able to reduce the viewer and stream size by going to widescreen, and the viewing experience was actually better.

Server Resources

untitled-8.jpguntitled-7.jpguntitled-6.jpguntitled-5.jpguntitled-4.jpguntitled-3.jpguntitled-2.jpguntitled-1.jpg

that’s all folks

yes i know there’s more than goes into live streams, but this is just me giving a brief overview from an experience. hopefully this helps someone, and wasn’t just good intent, bad execution :)

Sid

If you’d like to see the event streams recorded check out : http://www.iequine.com/

off topic: frustrated and tired

24 Oct

**this is a rant, sorry but needed to vent**

so I’m bout to doze off when i get a message on skype. A clients dv has been locked up since 4pm. its now 11:20. I have no idea what happened, hell I’ve only been on it 3 times. but come on its an extreme thats only running rails and this is the same server mt lost the data on.

why would it now be maxed out cpu and memory wise when it handled the rails site for 2 weeks, and was able to take a licking from fms and keep going….

no friggin clue, but i may start recommending someone else if this becomes a trend :(

this is the first bad week in over 6 years of hosting with them. I still have faith, but can’t really defend them when the clients pissed and right.

I guess I’m just frustrated. some of the issues have nothing to do with mt. but all the client sees is the mt issues, regardless of it there were video issues on location, or brief communacation issues with the team, or hell me not seeing that they would get the traffic they got and getting a cdn to begin with. All they see is that we had an issue it got fixed, and then they feel mt dropped a major ball. to top it off the not migrating the database thing completely broke the camels back. Now their borderline back to windows streaming.

this was the clients first time using flash instead of windows streaming. I’m frustrated and sad to see them go back to windows, even though this broadcast has gotten the highest reviews from users than any other they’ve done. many of which have out right said : the quality and ease of this years broadcast was better than the vegas event, etc. It’s like you show someone a better way because you want to help them. But i also understand that to them windows is easy and familiar. not to mention their pissed off because our feed is now branded with a competitor, and theres nothing we can do.

In the end i guess the blame rests on me. They had on idea what type of traffic they would get, instead of starting small and having the nitro as a back up, I should have just put them on the cdn to jump, might have cost more from start, but then we’d never have had these issues with mt.

its been a long day, hopefully i’ll have some fun code posts tomorrow

where do they run when the media temples gone

24 Oct

Apparently more than a few people are running to a small orange from media temples grid hosting. I’m checking them out now, might get a server to play with. will post if i do.
A Small Orange – Virtual Private Servers – Plans

Powered by ScribeFire.

guess whose losing a contract

23 Oct

well I’m now losing the contract to create a webcasting system. I still have no friggin clue why media temple didn’t migrate the data back to the server when they migrated it back to an extreme.

luckily the events are backed up automatically. so now 3000+ users lost, a shitload of hate mail from those users, and a dropped contract. awesome.