charity

mongodata3:SECONDARY> c.find({ $query: {ppoi_0: { $nearSphere: [ -74.5287, 40.1301 ], $maxDistance: 0.02601798524805497 } , deviceType: "ios", channels: { $in: [ "featured_coupons", "special_sales" ] }, appVersion: { $in: [ "3.0", "3.0.1", "3.1", "3.5" ] } }}).explain()
{
        "cursor" : "BasicCursor",
        "isMultiKey" : false,
        "n" : 0,
        "nscannedObjects" : 6484024,
        "nscanned" : 6484024,
        "nscannedObjectsAllPlans" : 6484024,
        "nscannedAllPlans" : 6484024,
        "scanAndOrder" : false,

charity

Failure is Not Optional -- it will happen

http://highscalability.com/blog/2010/10/15/troubles-with-sharding-what-can-we-learn-from-the-foursquare.html
https://web.archive.org/web/20110209190434/http://blog.foursquare.com/2010/10/05/so-that-was-a-bummer/

https://www.joyent.com/blog/postmortem-for-outage-of-us-east-1-may-27-2014

https://aws.amazon.com/message/5467D2/

http://perfcap.blogspot.com/2012/11/cloud-outage-reports.html

charity

Introducing a new series: Post-Mortem Book Reports

Dear fellow systems engineers,

Take a moment and think about the past few years in systems outages and public post mortems. 

What were your favorite outages?  What are the post-mortems that you read that stick with you, months or years or years and years later?  What did you learn from them?

If you are in AWS us-east-1, you probably think back to the Christmas Eve outage of 2013 or the long string of EBS outages.  If you were an early user of mongo sharding, I'm betting the 4sq mongo outage is etched into your brain.  If you run physical data centers and run your own networking or experience lots of DDOS attempts, GitHub post mortems are probably high on your list.

charity

#!/bin/bash -xe
# requires jq 1.5 (or at least > 1.3) and kafkacat

PATH=$PATH:/usr/lib/kafka/bin
topic="hound-staging.retriever-mutation"

which kafkacat || echo 'no kafkacat found, bye!' && exit 1
which jq || echo 'no jq found, bye!' && exit 1

# make sure jq is v 1.5

charity

PRODUCTION root@kafka-6c3f65f0:~# kafkacat -L -b localhost:9092 
Metadata for all topics (from broker -1: localhost:9092/bootstrap):
 4 brokers:
  broker 1003 at ip-10-0-246-113.ec2.internal:9092
  broker 1004 at ip-10-0-190-45.ec2.internal:9092
  broker 1009 at ip-10-0-213-10.ec2.internal:9092
  broker 1001 at ip-10-0-148-52.ec2.internal:9092
 4 topics:
  topic "hound-dogfood.retriever-mutation" with 1 partitions:
    partition 0, leader 1001, replicas: 1001, isrs: 1001

charity

Introducing a new series: Post-Mortem Book Reports

Dear fellow systems engineers,

Take a moment and think about the past few years in systems outages and public 
post mortems.

What were your favorite outages?  What are the post-mortems that you read that 
stick with you, months or years or years and years later?  What did you learn
from them?

charity

$ for svc in $(aws list 3>&1 1>&2 2>&3 3>&- | sed -e '1,7d' |sed -e 's/\|//g') ; do aws $svc describe-account-attributes 2>/dev/null || echo "not supported for $svc" ; done

charity

## declare all the env-specific variables that are defined in *.tfvars

variable "env" { } 
variable "name" { } 
variable "size" { } 

variable "cidr" { } 
variable "instance_type" { } 
variable "kafka_instance_type" { } 
variable "retriever_instance_type" { } 

charity

### these variables should be the same across all environments
# your aws secret key and access key should be in your env variables
provider "aws" {
    region = "us-east-1"
}

variable "tf_s3_bucket" { default = "hound-terraform-state" }
variable "master_state_file" { default = "base.tfstate" }
variable "prod_state_file" { default = "production.tfstate" } # TODO: make init.sh use these variables
variable "staging_state_file" { default = "staging.tfstate" }

charity

# file name terraform/modules/aws_vpc/vpc.tf
# first create the VPC.
# Prefix resources with var.name so we can have many environments trivially

resource "aws_vpc" "mod" {
  cidr_block = "${var.cidr}"
  enable_dns_hostnames = "${var.enable_dns_hostnames}"
  enable_dns_support = "${var.enable_dns_support}"
  tags { 
    Name = "${var.env}_vpc"