Nova Resource:Deployment-prep/Help

From Wikitech
Jump to navigation Jump to search

This page describes the essential bits for people with shell access to Beta Cluster.

Icinga Overview


If you are setting up a new instance, remember that Beta Cluster has its own puppet master. See Help:Standalone puppetmaster for more details.

If you are deleting an instance, follow the Decommission an instance checklist.


  • deployment-deploy01: maintenance script host
  • deployment-apacheXX : apaches, only ssh there in order to reload apache
  • deployment-cache-XXX : squid caches, where XXX is a role, each one should have a different public IP address
    • deployment-cache-bits : to emulate
    • deployment-cache-uploadXX : upload squid
  • deployment-tin : this is instance you want to ssh to, contains all stuff you need and all maintenance should be done there, even apache runs here. Similar to production deployment hosts (see deployment.eqiad.wmnet).
  • deployment-fluorine : host udp2log, the Mediawiki log receiver. No need to log there, logs are shared in /home/wikipedia/logs
  • deployment-videoscalerXX  : box which should do the video resizing
  • deployment-jobrunnerXX : boxes running MediaWiki asynchronous jobs. No web hits. Logs in shared /home/wikipedia/logs/runJobs.log
  • deployment-logstash1 : Logstash log aggregator
  • deployment-memcXX : memcached
  • deployment-pdf01 : PDF rendering
  • deployment-salt : Local Salt master and Puppet master
  • deployment-dbX : main sql server, isn't necessary to ever ssh there, unless you do dba stuff, or mysql is down


Configuration lives in /usr/local/apache/common/wmf-config - this is a shared directory

/usr/local/apache/common is a git repository, please commit all stuff you change

Always log all stuff to SAL by posting !log message in #wikimedia-releng connect!