1.7.6 connectivity requirements

• Collectors (optionally) to HBase (Thrift:9090)
• Crash mover to HBase (Thrift:9090)
• Monitor to HBase (Thrift:9090), PG (via Zeus:5432)
• Processor to HBase (Thrift:9090), PG (via Zeus:5432)
• Processor pulls symbols for mdsw from an NFS mount; no firewall rule needed
• PostgreSQL has other things talk to it only
• Middleware to HBase (Thrift:9090), PG (via Zeus:5432), memcache(11211)
• Webapp to PG (via Zeus:5432), middleware (80, probably via Zeus?), memcache(11211)
• Memcache to itself [jabba supply details here]
• Cron jobs (on admin node) to:
  • Bugzilla cron connects to bugzilla on 80
  • builds cron connects to builds FTP server (20) and PG
  • Partition cron connects to PG
  • topsites cron connects to alexa(80) and PG
  • dailyurl connects to NFS (for CSVs)
  • all other crons just connect to PG
• Thrift has other things talk to it only
• Hbase to Monitor (port?)
• Zeus to PG (6432), webapp (80), mware (80?)
• MPT-VPN to all PHX boxes (ssh:443) [jabba: or should this be to admin, and then ssh from admin->all other boxes]
• Nagios/ganglia? [jabba to fill in]
• Metrics ADU updates connect to PostgreSQL [daniel to supply machine name]

1.7.5 requirements

• In a handy, but hand-drawn diagram

Route Requirements, in Detail

PostgreSQL Connections

"PostgreSQL" below refers to BOTH master01 and master02

• Zeus --> PostgreSQL port 5432 (working)
• Zeus --> PostgreSQL port 6432 (working)
• Monitors --> Zeus 5432 (not tested)
• Processors --> Zeus 5432 (not tested)
• Middleware --> Zeus 6432 (not tested)
• Webapp --> Zeus 6432 (not tested)
• Cron Master --> Zeus 5432 (not tested)
• Memcache Server(s) --> Zeus 6432 (not tested)
• Metrics Server --> Zeus 5432 (not tested)
• Ganglia Server --> Zeus 5432 (not tested)
• Ganglia Server --> PostgreSQL 8849 (ganglia agent port) (not tested)
• Nagios Server --> Zeus 5432 (not tested)
• Nagios Server --> Zeus 6432 (not tested)
• Nagios Server --> PostgreSQL 6432 (not tested)
• Nagios Server --> PostgreSQL 5432 (not tested)
• Nagios Server --> PostgreSQL Nagios port (5667?) (not tested)
• PostgreSQL Master01 <--> Master02: SSH, Rysnc, 5432 (working)

Migration routes (all are currently working, do not disable until after migration)

• Old PostgreSQL server <--> new Master01 SSH (working)
• Old PostgreSQL server <--> new Master01 5432 (working)

Contingency Routes: these are routes to allow us to continue processing crashes in the event that a failure with Zeus renders it inoperable for several hours.

• Monitors --> PostgreSQL 5432
• Processors --> PostgreSQL 5432