Better Uptime Than Ever
Thanks in large part to New Relic, FIT Radio now delivers better uptime than ever. "Our uptime is basically 99.999%," says Ringersma. "Prior to New Relic, we were maybe at 98%. Customers notice that. Best of all, we no longer rely on users to perform site monitoring for us — and that goes a long way toward improving customer satisfaction."
Read the Full Case Study
To find out how New Relic helped FIT Radio cut its troubleshooting …
- Lessons Learned: Kiwi lean startup + Australia next
- High Scalability: Stuff The Internet Says On Scalability For April 13, 2012
- Jerrett Taylor: Reality Check, Mr. Harper.
- Luigi Montanez: Stop Using the text-indent:-9999px CSS Trick
- Rails Inside: How a Realtime SMS Questions Service Scaled To Super Bowl Traffic with Rails
With a self-imposed goal of 99.9% uptime, the company makes specific plans for March's huge traffic spike. New Relic's application performance monitoring tool is a critical part of these efforts. Bleacher Report depends on New Relic to achieve its performance goals and meet unexpected demand as traffic surges around breaking news stories.
" New Relic is the core of our performance monitoring with a large screen up in our development and operations room to monitor our …
- Comments on: blog: Link dump for September 29th
- Effectif Development: 3 Unix commands for finding performance problems
- RailsBridge News: Do One Thing for Rails 3 on January 16th and 17th
- Engine Yard Developer Blog: Engine Yard Portland is open for business!
- Ruby Inside: Check Out The Companies That Make Ruby Inside Possible
- paperplanes: The Virtues of Monitoring
Uptime numbers for the last month: At the top, you'll see uptime numbers for the previous month. This provides an at-a-glance answer to the question of "How stable has Heroku been lately?" Our numbers here haven't been as good as we'd like in the last few months, but by posting them publicly here we intend to create the transparency and accountability that will help drive us to improve. After all, you make what you measure .
Timeline: Most status sites, including …
- DigitalHobbit: Rails in the Cloud: AWS, Heroku, and Morph
- RubyLearning Blog: Getting started with Heroku
- gnuu.org: Why are the Numbers So Whack? Plausible Explanations for Crazy Gem Download Counts
- techno weenie: Get your Github news feed in Tweetie
- Envy Labs: Try Ruby Rewrite and Redesign
- Global Nerdy: At the Ottawa Rails 3.1 Hackfest
Check whether the Uptime variable is too small, indicating the server has been restarted
Check whether Threads connected is approaching max_connections, which can be inconvenient to solve in some environments
However, there's no reason not to use the plugin for other purposes if you want. Here are some of the things you could do:
Compare a counter to a threshold
Compare a counter to a variable or another counter
Add, subtract, multiply, and divide counters or variables
Transform …
- Opscode Blog: Opscode Unveils Private Chef for the Enterprise
…much more serious. You can use the pmp-check-mysql-status plugin to alert when the server's Uptime variable is too small.
The server is approaching max_connections. Overflowing the max_connections setting is a bit like a lack of disk space: it is possible for it to be slowly approached over time, and then to begin causing problems when connections are rejected. Similarly to deadlocks and timeouts, applications often don't handle or log this error appropriately. The pmp-check-mysql-status …
- Signal vs. Noise: Nuts & Bolts: Database Servers
- New Relic: How do you move more than 6000 customers in 2 hours? (Hint: plan ahead)
- paperplanes: MongoDB, Data Durability and Improvements coming in 1.8
- Engine Yard Developer Blog: MySQL 5.5 and 5.1 Support in Alpha
- Journal: Faye integration into a rails app (+ testing!)
- PeepCode Products: Meet PostgreSQL
Scale The Amazon whitepaper on running MongoDB on AWS is full of good info. Nothing earth shattering, just not anything you want to learn the trial & error way.
Yields Interesting read about co-routines, generators and JavaScript: Why coroutines won't work on the web .
Worth repeating Life would be simpler if GET, PUT and DELETE were all, always, idempotent .
Uptime Turns out morning people and nightowls show different brain function :
PID : 26049 Sessions: 20 Processed: 2977 Uptime: 2h 53m 55s
There is a 20 session backlog for the process above: it looks like we need to increase the number of Passenger processes.
Tuning Passenger
As I mentioned earlier, Passenger is likely to be the biggest consumer of memory in your web stack. It requires some special attention. Take a look at our previous post: Production Rails Tuning with Passenger: PassengerMaxProcesses for instructions on tuning Passenger. …
- High Scalability: Scaling DISQUS to 75 Million Comments and 17,000 RPS
- Rubynaut: Scaling a Rails app (Part 1 of 3)
- Litany Against Fear: RailsConf 2009 Guitar Hero Behind The Music
- elc technologies - Blog : Ruby on Rails, iPhone, Flash, Agile, Interactive Design: Ruby Daemons and Angels
- Luke Melia: Running REE at EngineYard
- Opscode Blog: Chef on Steroids (Demo Tonight) at LOPSALA
…false otherwise. It can be used for example if you want to know how long the server has been running. Uptime is a Unix/ Linux command to do this: >> system "uptime" 15:39:41 up 5:11, 3 users, load average: 0.13, 0.27, 0.35 => true
Another example is to get the amount of free space on your hard disks. In this case you could use df (disk free) to display the amount of available disk space: >> system "df -h" Filesystem Size Used Avail Use% Mounted …
Key cache miss rate = Key reads / Uptime
Note the conspicuous absence of Key read requests in the formula. The number of requests is absolutely irrelevant -- who cares how often the key is requested? What's relevant is that our assumed connection between Key reads and random I/ Os means that Key reads/ Uptime is assumed to be the same as "random I/ Os per second."
And now, I would finally like to show you something partially useful you can do with Key reads: