bpot

curl http://localhost:6060/debug/pprof/goroutine?debug=2 | tee goroutine.txt | grep -A3 goroutine | grep ":" | sort | uniq -c | sort -rn | head

bpot

Keybase proof

I hereby claim:

• I am bpot on github.
• I am bpot (https://keybase.io/bpot) on keybase.
• I have a public key whose fingerprint is A287 568E 79FA 759A 1300 8FC7 8B3D EF7D 9D8F B781

To claim this, I am signing this object:

bpot

require 'ffi'

module Block
  extend FFI::Library

  ffi_lib FFI::Library::LIBC

  attach_function :sleep, [:uint], :void
end

bpot

[email protected]

bpot

Kafka failures when sending to a single topic:

• We can't connect to any brokers
• We connect to a broker but cluster metadata fetch fails
• Metadata doesn't know about topic
• Metadata has no leader for that partition we want to send to
• There is a leader for the partition but we can't connect to it
• There is a leader for the partition and we sent the request but we timedout or the connection broke (unclear if success)
• There is a leader for the partition and we sent the request but the broker returned an error
• There is a leader for the partition, required_acks > 1 and we sent the request but the leader didn't receive acks in time.

bpot

[email protected]
[email protected]
[email protected]

bpot

Dynotype

A write efficient implementation of large (consistent) sets on DynamoDB.

Why?

DynamoDB has built-in support for sets but if your sets are large (>1000 members) you quickly run into two issues:

1. When calculating the cost of an item update AWS charges you based on the size of the item and not the size of the update. This makes even small additions to large sets expensive.

bpot

# Attempt to reproduce SSL_shutdown() hang. http://bugs.ruby-lang.org/issues/7584
require 'net/http'
require 'uri'

20.times do  
  Thread.new {
    loop do
      # Sorry google!
      uri = URI.parse("https://www.google.com/")
      http = Net::HTTP.new(uri.host, uri.port)

bpot

#Tuesday (3/12)

Mohawk (Outside)

• 1:00pm - Lemuria

Valhalla

• 5:00pm - Japanther

bpot

a