repertoire-repertoire-devtools 0.0.1 → 0.0.2

data/README.txt

@@ -0,0 +1,12 @@
+This gem contains tools for use in configuring Hyperstudio servers and workstations.  Currently it contains:
+
+Process management tools:
+
+- merb_persist:             adaptation of Apple's mongrel_rails_persist launchd helper, for merb
+- postgresql_crontab_sweep: crontab like tasks for postgresql (esp. for full text indexing)
+
+All are ruby applications; see individual files for more information.  To install:
+
+sudo gem install repertoire-repertoire-devtools --source http://gems.github.com
+
+[ repetition in the gem name is not a typo - it's a side effect of github's gem naming system ]

data/bin/postgresql_crontab_sweep

@@ -5,6 +5,8 @@
 require "rubygems"
 require "pg"
 
+require 'benchmark'
+
 #
 # Basic support for crontab-type tasks within PostgreSQL
 #
@@ -19,33 +21,36 @@ require "pg"
 #   Before running, create the crontab table as follows:
 #
 #     CREATE TABLE crontab(id SERIAL PRIMARY KEY,
-#                   notice TEXT NOT NULL,
-#                    task TEXT NOT NULL, 
-#                    interval INTERVAL NOT NULL, 
-#                  last_updated_at TIMESTAMP WITHOUT TIME ZONE NOT NULL DEFAULT 'epoch',
-#                   failing BOOLEAN NOT NULL DEFAULT false,
-#                   message TEXT);
+#                          role TEXT,
+#                          notice TEXT NOT NULL,
+#                          task TEXT NOT NULL, 
+#                          interval INTERVAL NOT NULL, 
+#                          last_updated_at TIMESTAMP WITHOUT TIME ZONE NOT NULL DEFAULT 'epoch',
+#                          time_elapsed INTERVAL,
+#                          failing BOOLEAN NOT NULL DEFAULT false,
+#                          message TEXT);
 #
 #   This script makes a single sweep through the crontab list, executing pending tasks.  To use,
 #     run as a cron script (standard UNIX), or launchd task (OS X).
 #
-#   If you would like the script to set up the database table and a basic launchd task, you can use
-#     the following shortcut:
-#   `sudo postgresql_crontab_sweep install <database-name>`
-#     on the command line (OS X only).  This will generate the relevant files, tables, and start the 
-#     update polling process.
+#   If you would like the script to set up the database table and a OS X launchd task, you can use
+#     the 'install' shortcut.  For more information, see the help page available at 'postgresql_crontab_sweep -h'
 #
 #   Each task is executed in a separate transaction, so they fail or succeed separately.  If the
 #     task commits, its timestamp is updated.  If it aborts, last_updated_at will show the most
 #     recent successful run, and the failing and message fields will be set.
 #
 #   Parameters:  -d <database> -p <port> -h <host> -u <user> -p <password>  [ all are optional ]
-#           or:  install <database>
+#           or:  install <database> [ -i <interval in seconds> ]
 #
 #   Because this script should only be run locally to the database, it expects postgres to accept
 #     your user without a password prompt (see pg_hba.conf to configure).  Usually this user will
-#     be 'postgres', the db admin role.  When writing tasks, keep in mind that this user will
-#     execute them.  You may need to qualify tables in separate schemas, and grant access to appropriately.
+#     be 'postgres', the db admin role.  
+#
+#   When writing tasks, keep in mind that this user will execute them, unless the 'role' column has been
+#     filled.  In this case, a SET ROLE <role> command will execute first, which allows the task to be
+#     written as though the provided role were executing it (most importantly, you can assume user's schema
+#     as default.  You will need to grant this role to the superuser before this feature can be used, however.
 #
 #   Christopher York 12/8/2008
 #
@@ -57,7 +62,7 @@ def run(dbparams)
   forking  = false
 
   # get the pending database cron tasks
-  pending  = cron_conn.exec("SELECT id, notice, task, interval, last_updated_at FROM crontab WHERE last_updated_at + interval < now();")
+  pending  = cron_conn.exec("SELECT * FROM crontab WHERE last_updated_at + interval < now();")
 
   pending.each do |crontask|
     # fork a sub-process for each task, with its own database connection
@@ -68,8 +73,12 @@ def run(dbparams)
         # all postgresql cron tasks take place in their own transaction (which is the reason for this script - not possible in plpgsql)
         task_conn.transaction do 
           puts crontask[:notice]
-          task_conn.exec("UPDATE crontab SET last_updated_at = now(), failing = false, message = NULL WHERE id = #{ crontask['id'] }")
-          task_conn.exec(crontask['task'])
+          task_conn.exec("SET ROLE #{ crontask['role'] }") if crontask['role']
+          time_elapsed = Benchmark.realtime do
+            task_conn.exec(crontask['task'])
+          end
+          task_conn.exec("RESET ROLE")
+          task_conn.exec("UPDATE crontab SET last_updated_at = now(), time_elapsed = '#{ time_elapsed }', failing = false, message = NULL WHERE id = #{ crontask['id'] }")
         end
       rescue PGError => e  
         STDERR.puts "ERROR!  #{ Time.now } - #{ e.message }"
@@ -84,50 +93,74 @@ def run(dbparams)
   Process.wait if forking   # avoid zombies by waiting for child processes to finish
 end
 
-def install(dbparams)
-  launchd_plist = <<-END_OF_PLIST
-<?xml version="1.0" encoding="UTF-8"?>
-<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
-<plist version="1.0">
-<dict>
-  <key>Label</key>
-  <string>edu.mit.hyperstudio.postgresql.crontab.plist</string>
-  <key>OnDemand</key>
-  <false/>
-  <key>ProgramArguments</key>
-  <array>
-    <string>/usr/bin/postgresql_crontab_sweep</string>
-    <string>-d</string>
-    <string>#{dbparams['dbname']}</string>
-  </array>
-  <key>StartCalendarInterval</key>
-  <dict>
-    <key>StartInterval</key>
-    <integer>5</integer>
-  </dict>
-</dict>
-</plist>
-END_OF_PLIST
-
-  create_table = <<-END_OF_CREATE_TABLE
+def install(dbparams, script_args, sweep_interval)
+  require 'osx/cocoa'
+  include OSX
+
+  create_table_stmt = <<-END_OF_CREATE_TABLE
 CREATE TABLE crontab(id SERIAL PRIMARY KEY,
                      notice TEXT NOT NULL,
+                     role TEXT,
                      task TEXT NOT NULL, 
                      interval INTERVAL NOT NULL, 
                      last_updated_at TIMESTAMP WITHOUT TIME ZONE NOT NULL DEFAULT 'epoch',
+                     time_elapsed INTERVAL,
                      failing BOOLEAN NOT NULL DEFAULT false,
                      message TEXT);
 END_OF_CREATE_TABLE
 
-  File.open('/Library/LaunchDaemons/edu.mit.hyperstudio.postgresql.crontab.plist', 'w') { |f| f.write(launchd_plist) }
-  
+  # create database crontab table
+  puts "Creating crontab table in database #{ dbparams[:dbname] }"
   conn = PGconn.open(dbparams)
-  conn.exec(create_table)
+  begin
+    conn.exec(create_table_stmt)
+  rescue PGError => e  
+    STDERR.puts "WARNING: #{ Time.now } - #{ e.message }"
+  end
   conn.close
+
+  # create and install the crontab launchd task
+  script_path = File.expand_path(File.dirname(__FILE__) + '/' + File.basename(__FILE__))
+  plist_label = "edu.mit.hyperstudio.postgresql.crontab"
+  plist_path  = "/Library/LaunchDaemons/#{plist_label}.plist"
+
+  if ! test(?d,File.dirname(plist_path))
+  	raise unless system('/bin/mkdir', '-m', '0755', plist_path)
+  	raise unless system('/usr/sbin/chown', 'root:wheel', plist_path)
+  end
+
+  h = {}
+  h['Label'] = plist_label
+  h['OnDemand'] = false
+  h['ProgramArguments'] = [script_path, script_args.to_a].flatten.compact
+  h['StartCalendarInterval'] = { 'StartInterval' => sweep_interval }
+
+  data = NSPropertyListSerialization.dataFromPropertyList_format_errorDescription(h, NSPropertyListXMLFormat_v1_0, nil)
+  plist = NSString.alloc.initWithData_encoding(data, NSUTF8StringEncoding)
+  File.open(plist_path, 'w') { |f| f.puts plist }
+
+  `/bin/launchctl unload #{plist_path} >/dev/null 2>&1`
+  raise unless system('/bin/launchctl', 'load', plist_path)
+  puts "Loaded #{plist_path}\nPostgreSQL crontab sweeper should persist across reboot."
+  puts "To stop: /bin/launchctl unload -w #{plist_path}"
+  
+  exit
 end
 
 
-# parse params, with defaults
+# parse params, with defaults; handle usage
+
+	usage = <<EOU
+Usage: #{File.basename($0)} -h | [-d <database>] [-p <port> ] [-u <user>] [-P <password>] [install]
+This is a sweep script for Repertoire's PostgreSQL crontab support.  Run once to sweep through the crontab table,
+executing each task in a separate transaction/process.  Or run with the 'install' parameter to create the
+crontab table and start periodic launchd task to run it.  See in-script comments for more details.
+EOU
+
+if ARGV.empty? || ARGV.include?('-h')
+  $stderr.puts usage
+	exit
+end
 
 opts = {
   :dbname   => 'hyperstudio_development',
@@ -135,16 +168,26 @@ opts = {
   :port     => 5432,
   :host     => 'localhost'
 }
+args = {}
+installing = false
+sweep_interval = 5
 
 while opt = ARGV.shift
 	case opt
-	when '-d' then opts['dbname'] = ARGV.shift
-	when '-p' then opts['port'] = ARGV.shfit
-	when '-u' then opts['user'] = ARGV.shift
-	when '-p' then opts['password'] = ARGV.shift
-  when 'install' then install(opts) && exit
+	when '-d' then opts[:dbname]   = args['-d'] = ARGV.shift
+	when '-p' then opts[:port]     = args['-p'] = ARGV.shift
+	when '-u' then opts[:user]     = args['-u'] = ARGV.shift
+	when '-P' then opts[:password] = args['-P'] = ARGV.shift
+  when 'install' then installing = true
+  when '-i' then sweep_interval = ARGV.shift
+  else raise "Unkown argument #{opt}"
   end
 end
 
-run(opts)
+if installing
+  install(opts, args, sweep_interval)
+else
+  run(opts)
+end
+
 exit

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification 
 name: repertoire-repertoire-devtools
 version: !ruby/object:Gem::Version 
-  version: 0.0.1
+  version: 0.0.2
 platform: ruby
 authors: 
 - Christopher York
@@ -9,7 +9,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2008-12-15 00:00:00 -08:00
+date: 2009-01-02 00:00:00 -08:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -17,9 +17,9 @@ dependencies:
   version_requirement: 
   version_requirements: !ruby/object:Gem::Requirement 
     requirements: 
-    - - ">"
+    - - ">="
       - !ruby/object:Gem::Version 
-        version: 0.0.0
+        version: "0"
     version: 
 description: Development and server tools for use with Hyperstudio Repertoire
 email: yorkc@mit.edu