search

nodeconf / EU-CFP

Call for participation for NodeConf.eu 2015
10 stars 6 forks source link

The magic dump - when something in node goes wrong #3

Open lucamaraschi opened 9 years ago

lucamaraschi commented 9 years ago

3AM, production down. The door of the war room opens and the walls are covered by screens showing the terrible scene of machines going in memory overflow and dying. Which strategy to use?Where to start?Who is responsible of this crisis?Panic in the room. Check the logs, read all the informations in Kibana...nothing...we are all in a pile of s**t. Fortunately this was only a nightmare but what if it becomes reality?We all know that our node services are going to crash, that our code is somehow leaking somewhere but do we know how to start investigating into the problems?Do we know how to save time and look for the problem directly into the source? A real life learning and list of practices/processes that I adopted when I was in that war room.

The goal for the attendees of this talk is to learn and discover how to trace and analyse a node process: a journey which will take them from DTrace to post-mortem analysis with MDB, using a real case scenario and following the path that I did while searching for that (before) unknown reason of suddenly crashing our node services.

mikeal commented 8 years ago

heya, we need to contact you about your talk. can you send your email address to mikeal.rogers@gmail.com :)