gaworkflow/automation/spoolwatcher.py

   1 #!/usr/bin/env python
   2 import logging
   3 import os
   4 import re
   5 import sys
   6 import time
   7 #import glob
   8
   9 from gaworkflow.util import mount
  10
  11 # this uses pyinotify
  12 import pyinotify
  13 from pyinotify import EventsCodes
  14
  15 from benderjab import rpc
  16
  17
  18 class WatcherEvents(object):
  19     # two events need to be tracked
  20     # one to send startCopy
  21     # one to send OMG its broken
  22     # OMG its broken needs to stop when we've seen enough
  23     #  cycles
  24     # this should be per runfolder.
  25     # read the xml files
  26     def __init__(self):
  27         pass
  28
  29
  30 class Handler(pyinotify.ProcessEvent):
  31     def __init__(self, watchmanager, bot):
  32         self.last_event_time = None
  33         self.watchmanager = watchmanager
  34         self.bot = bot
  35
  36     def process_IN_CREATE(self, event):
  37         self.last_event_time = time.time()
  38         msg = "Create: %s" %  os.path.join(event.path, event.name)
  39         if event.name.lower() == "run.completed":
  40             try:
  41                 self.bot.sequencingFinished(event.path)
  42             except IOError, e:
  43                 logging.error("Couldn't send sequencingFinished")
  44         logging.debug(msg)
  45
  46     def process_IN_DELETE(self, event):
  47         logging.debug("Remove: %s" %  os.path.join(event.path, event.name))
  48
  49     def process_IN_UNMOUNT(self, event):
  50         pathname = os.path.join(event.path, event.name)
  51         logging.debug("IN_UNMOUNT: %s" % (pathname,))
  52         self.bot.unmount_watch()
  53
  54 class SpoolWatcher(rpc.XmlRpcBot):
  55     """
  56     Watch a directory and send a message when another process is done writing.
  57
  58     This monitors a directory tree using inotify (linux specific) and
  59     after some files having been written will send a message after <timeout>
  60     seconds of no file writing.
  61
  62     (Basically when the solexa machine finishes dumping a round of data
  63     this'll hopefully send out a message saying hey look theres data available
  64
  65     """
  66     # these params need to be in the config file
  67     # I wonder where I should put the documentation
  68     #:Parameters:
  69     #    `watchdir` - which directory tree to monitor for modifications
  70     #    `profile` - specify which .gaworkflow profile to use
  71     #    `write_timeout` - how many seconds to wait for writes to finish to
  72     #                      the spool
  73     #    `notify_timeout` - how often to timeout from notify
  74
  75     def __init__(self, section=None, configfile=None):
  76         #if configfile is None:
  77         #    self.configfile = "~/.gaworkflow"
  78         super(SpoolWatcher, self).__init__(section, configfile)
  79
  80         self.cfg['watchdir'] = None
  81         self.cfg['write_timeout'] = 10
  82         self.cfg['notify_users'] = None
  83         self.cfg['notify_runner'] = None
  84
  85         self.notify_timeout = 0.001
  86         self.wm = pyinotify.WatchManager()
  87         self.handler = Handler(self.wm, self)
  88         self.notifier = pyinotify.Notifier(self.wm, self.handler)
  89         self.wdd = None
  90         self.mount_point = None
  91         self.mounted = True
  92
  93         self.notify_users = None
  94         self.notify_runner = None
  95
  96         self.eventTasks.append(self.process_notify)
  97
  98     def read_config(self, section=None, configfile=None):
  99         super(SpoolWatcher, self).read_config(section, configfile)
 100
 101         self.watch_dir = self._check_required_option('watchdir')
 102         self.write_timeout = int(self.cfg['write_timeout'])
 103
 104         self.notify_users = self._parse_user_list(self.cfg['notify_users'])
 105         try:
 106           self.notify_runner = \
 107              self._parse_user_list(self.cfg['notify_runner'],
 108                                    require_resource=True)
 109         except bot.JIDMissingResource:
 110             msg = 'need a full jabber ID + resource for xml-rpc destinations'
 111             logging.FATAL(msg)
 112             raise bot.JIDMissingResource(msg)
 113
 114     def add_watch(self, watchdir=None):
 115         """
 116         start watching watchdir or self.watch_dir
 117         we're currently limited to watching one directory tree.
 118         """
 119         # the one tree limit is mostly because self.wdd is a single item
 120         # but managing it as a list might be a bit more annoying
 121         if watchdir is None:
 122             watchdir = self.watch_dir
 123         logging.info("Watching:"+str(watchdir))
 124
 125         self.mount_point = mount.find_mount_point_for(watchdir)
 126
 127         mask = EventsCodes.IN_CREATE | EventsCodes.IN_UNMOUNT
 128         # rec traverses the tree and adds all the directories that are there
 129         # at the start.
 130         # auto_add will add in new directories as they are created
 131         self.wdd = self.wm.add_watch(watchdir, mask, rec=True, auto_add=True)
 132
 133     def unmount_watch(self):
 134         if self.wdd is not None:
 135             self.wm.rm_watch(self.wdd.values())
 136             self.wdd = None
 137             self.mounted = False
 138
 139     def process_notify(self, *args):
 140         # process the queue of events as explained above
 141         self.notifier.process_events()
 142         #check events waits timeout
 143         if self.notifier.check_events(self.notify_timeout):
 144             # read notified events and enqeue them
 145             self.notifier.read_events()
 146             # should we do something?
 147         # has something happened?
 148         last_event_time = self.handler.last_event_time
 149         if last_event_time is not None:
 150             time_delta = time.time() - last_event_time
 151             if time_delta > self.write_timeout:
 152                 self.startCopy()
 153                 self.handler.last_event_time = None
 154         # handle unmounted filesystems
 155         if not self.mounted:
 156             if mount.is_mounted(self.mount_point):
 157                 # we've been remounted. Huzzah!
 158                 # restart the watch
 159                 self.add_watch()
 160                 self.mounted = True
 161                 logging.info(
 162                     "%s was remounted, restarting watch" % \
 163                         (self.mount_point)
 164                 )
 165
 166     def _parser(self, msg, who):
 167         """
 168         Parse xmpp chat messages
 169         """
 170         help = u"I can send [copy] message, or squencer [finished]"
 171         if re.match(u"help", msg):
 172             reply = help
 173         elif re.match("copy", msg):
 174             self.startCopy()
 175             reply = u"sent copy message"
 176         elif re.match(u"finished", msg):
 177             words = msg.split()
 178             if len(words) == 2:
 179                 self.sequencingFinished(words[1])
 180                 reply = u"sending sequencing finished for %s" % (words[1])
 181             else:
 182                 reply = u"need runfolder name"
 183         else:
 184             reply = u"I didn't understand '%s'" %(msg)
 185         return reply
 186
 187     def start(self, daemonize):
 188         """
 189         Start application
 190         """
 191         self.add_watch()
 192         super(SpoolWatcher, self).start(daemonize)
 193
 194     def stop(self):
 195         """
 196         shutdown application
 197         """
 198         # destroy the inotify's instance on this interrupt (stop monitoring)
 199         self.notifier.stop()
 200         super(SpoolWatcher, self).stop()
 201
 202     def startCopy(self):
 203         logging.debug("writes seem to have stopped")
 204         if self.notify_runner is not None:
 205             for r in self.notify_runner:
 206                 self.rpc_send(r, tuple(), 'startCopy')
 207
 208     def sequencingFinished(self, run_dir):
 209         # need to strip off self.watch_dir from rundir I suspect.
 210         logging.info("run.completed in " + str(run_dir))
 211         pattern = self.watch_dir
 212         if pattern[-1] != os.path.sep:
 213             pattern += os.path.sep
 214         stripped_run_dir = re.sub(pattern, "", run_dir)
 215         logging.debug("stripped to " + stripped_run_dir)
 216         if self.notify_users is not None:
 217             for u in self.notify_users:
 218                 self.send(u, 'Sequencing run %s finished' % (stripped_run_dir))
 219         if self.notify_runner is not None:
 220             for r in self.notify_runner:
 221                 self.rpc_send(r, (stripped_run_dir,), 'sequencingFinished')
 222
 223 def main(args=None):
 224     bot = SpoolWatcher()
 225     return bot.main(args)
 226
 227 if __name__ == "__main__":
 228     sys.exit(main(sys.argv[1:]))
 229