#DBHangOps 04/16/14 -- PLMCE Learnings and Defending your DBs!

Check out the recording below:

Hello everybody!

Join in #DBHangOps this Wednesday, April, 16, 2014 at 12:00pm pacific (19:00 GMT), to participate in the discussion about:

Be sure to check out the #DBHangOps twitter search, the @DBHangOps twitter feed, or this blog post to get a link for the google hangout on Wednesday!

See all of you on Wednesday!

Show notes

Particularly exciting/enjoyable talks
- ChatOps by Sam Lambert -- https://speakerdeck.com/samlambert/chatops-how-github-manages-mysql
- Devops @ Outbrain by Shlomi Noach -- http://www.slideshare.net/shlominoach/mysql-devops-at-outbrain
- Asynchronous client by Facebook
- GTID by facebook
Announcements
- Webscalesql -- http://webscalesql.org/
  - Pretty exciting to have a common upstream branch that major organizations are pushing changes and fixes into
Take aways
- Enabling GTIDs in an online cluster
  - Booking.com blogged about a small patch to mysql that allowed them to do it -- http://blog.booking.com/mysql-5.6-gtids-evaluation-and-online-migration.html

How do you defend your database from your own services?
- Killing connections based on thresholds
  - Long running queries
  - Long/large transactions
    - You can look at INFORMATION_SCHEMA.INNODB_LOCKS and INFORMATION_SCHEMA.INNODB_LOCK_WAITS to get information about locks by transaction.
  - Can use common_schema to get at information about transactions blocking each other -- http://common-schema.googlecode.com/svn/trunk/common_schema/doc/html/innodb_locked_transactions.html
What are critical metrics to measure?
- Slave availability for serving reads -- e.g. "7/10 slaves are available, we're okay"
  - Have slaves publish information to a service about their health (e.g. Zookeeper)
  - If a slave deems itself unhealthy, start throttling itself