EQNZ – crisis response, open source style – Brenda Wallace
- Started with a Trigger warning and “fucker”
- First thing posted – “I am okay” , one tweet, one facebook
- State of Scial Media
- Social media not as common, SMS king, not many smartphones
- Google Buzz, twitter, Facebook
- Multiple hashtags
- Questions people asked on social media
- Official info was under strain, websites down due to bad generators
- Crisis Commons
- Skype
- Free
- Multi-platform
- Txt based
- Battery Drain very bad
- Bad internet in Chc hard to use, no mobile, message reply for minutes on join
- Things pop up within an hour
- Pirate Pad
- Couch apps
- Wikis
- WordPress installs
- Short code 4000 for non-urgent help live by 5pm
- Volenteers processing the queue
- All telcos agree to coordinate their social media effort
- Civil defence didn’t have site ready and refused offers, people decided to do independantly
- Ushahidi instance setup
- Google setup people finder app
- Moved into ec2 cluther
- hackfest, including added mobile
- Some other Ushidis, in the end newspaper sites enbedded
- Council
- chc council wordpress for info
- Very slow and bad UI
- Hit very hard, old information from the previous earthquake
- staff under extreme pressure
- Civil Defence
- Official info only
- Falls over
- Caught by DDOS against another govt site
- Our reliability
- Never wen tdown
- contact and reassured some authorities
- After 24h . 78k page impressions
- Skype
- 100+ chatting. limitations
- IRC used by some but many no common enough
- Gap for something common. cross platform, easy to use
- Hashtag
- twitter to SMS notifications to add stuff to website
- Maps were a new thing
- None of the authorities knew them
- Council and DHB websites did not work on mobile and were not updating
- Government
- Govt officers didn’t talk – except NZ Geospacial office
- Meeting that some people attended
- Wrap up after 3 weeks
- Redirected website
- Anonymous copy of database
- Pragmatic
- Used closed source where we had too (eg skype)
- But easier with OS could quick to modify
- Closed source people could install webserver, use git, etc. Hard to use contributions
- Burned Bridges
- Better jobs with Gov agencies
- These days
- Tablets
- Would use EC2 again
- phones have low power mode
- more open street maps
collectd in dynamic environments – Florian Forster
- Started collectd in 2005
- Dynamic environments – Number and location of machines change frequently – VM or job management system
- NOTE: I use collectd so my notes are a little sparse here cause I knew most of it already
- Collects timeseries data, does one thing well. collectd.org
- agent runs on each host, plugins mostly in C for lots of things or exec plug to run random stuff.
- Read Plugins to get metrics from system metrics, applications, other weird stuff
- Write plugs – Graphite, RRD, Reimann, MongoDB
- Virtual machine Metrics
- libvirt plugin
- Various metrics, cpu, memory, swap, disk ops/bytes, network
- GenericJMX plugin – connects to JVM. memory and garbage collection, threads
- Network plugin
- sends and receives metric
- Effecient binary protocol. 50-100 byte UDP multicast/unicast protocol
- crypto available
- send, receive, forward packets
- Aggregation
- Often more useful for alerting
- Aggregation plugin
- Subscribes to metric
- aggregates and forwards
- Limitation, no state, eg medium, mean are missing
- only metrics with one value
- can be aggregated at any level
- eg instead of each CPU then total usage of all your CPUS
- Reimann
- Lots of filters and functions
- can aggregate, many otions
- Bosum
- Monitoring and alert language
- Storage
- Graphite
- OpenTSDB based on hadoop
- InfluxDB – understand collectd protocol native (and graphite).
- Vaultaire ( no collectd integration but… )
- New Dishboard – facette.io
