Totally re-work the way that jobfiles are done so there is more

information that can be used for other aspects of sending jobs.
New graphing output stuff with matplotlib.

util/pbs/job.py:
    Shuffle code around and create the JobDir class which encapsulates
    all of the functionality needed for making, organizing, and cleaning
    a job directory.
    Better status output
util/pbs/jobfile.py:
    Majory re-working of the jobfile code.
    A job file now consists of several objects that describe how
    jobs should be run, it includes information about checkpoints,
    and graphing.
util/pbs/send.py:
    use the new jobfile code.
    deal with the 15 character limit of pbs by truncating the name and
    using the raj hack.
util/stats/db.py:
    fix the __str__ function for nodes
    provide __getitem__ for the Database class
util/stats/stats.py:
    use the jobfile stuff to figure out what the proper naming
    and organziation of the graphs should be.
    move all output code to output.py, get rid of ploticus and use
    matplotlib

--HG--
rename : util/categories.py => util/stats/categories.py
extra : convert_revision : 0d793cbf6ad9492290e8ec875ce001c84095e1f7

1 files changed, 101 insertions, 38 deletions

diff --git a/util/pbs/job.py b/util/pbs/job.py
index f370862de..e2636c111 100755
--- a/util/pbs/job.py
+++ b/util/pbs/job.py
@@ -29,10 +29,21 @@
 #          Steve Reinhardt
 #          Ali Saidi
 
-import os, os.path, shutil, signal, socket, sys, time
+import os, os.path, shutil, signal, socket, sys
 from os import environ as env
 from os.path import join as joinpath, expanduser
 
+def date():
+    import time
+    return time.strftime('%a %b %e %H:%M:%S %Z %Y', time.localtime())
+
+def cleandir(dir):
+    for root, dirs, files in os.walk(dir, False):
+        for name in files:
+            os.remove(joinpath(root, name))
+        for name in dirs:
+            os.rmdir(joinpath(root, name))
+
 class rsync:
     def __init__(self):
         self.sudo = False
@@ -61,25 +72,76 @@ class rsync:
 
         return os.spawnvp(os.P_WAIT, args[0], args)
 
-def cleandir(dir):
-    for root, dirs, files in os.walk(dir, False):
-        for name in files:
-            os.remove(joinpath(root, name))
-        for name in dirs:
-            os.rmdir(joinpath(root, name))
+class JobDir(object):
+    def __init__(self, dir):
+        self.dir = dir
 
-def date():
-    return time.strftime('%a %b %e %H:%M:%S %Z %Y', time.localtime())
+    def file(self, filename):
+        return joinpath(self.dir, filename)
+
+    def create(self):
+        if os.path.exists(self.dir):
+            if not os.path.isdir(self.dir):
+                sys.exit('%s is not a directory.  Cannot build job' % self.dir)
+        else:
+            os.mkdir(self.dir)
+
+    def exists(self):
+        return os.path.isdir(self.dir)
+
+    def clean(self):
+        cleandir(self.dir)
+
+    def hasfile(self, filename):
+        return os.path.isfile(self.file(filename))
+
+    def echofile(self, filename, string):
+        filename = self.file(filename)
+        try:
+            f = file(filename, 'w')
+            print >>f, string
+            f.flush()
+            f.close()
+        except IOError,e:
+            sys.exit(e)
 
-def remfile(file):
-    if os.path.isfile(file):
-        os.unlink(file)
+    def rmfile(self, filename):
+        filename = self.file(filename)
+        if os.path.isfile(filename):
+            os.unlink(filename)
 
-def readval(filename):
-    file = open(filename, 'r')
-    value = file.readline().strip()
-    file.close()
-    return value
+    def readval(self, filename):
+        filename = self.file(filename)
+        f = file(filename, 'r')
+        value = f.readline().strip()
+        f.close()
+        return value
+
+    def setstatus(self, string):
+        filename = self.file('.status')
+        try:
+            f = file(filename, 'a')
+            print >>f, string
+            f.flush()
+            f.close()
+        except IOError,e:
+            sys.exit(e)
+
+    def getstatus(self):
+        filename = self.file('.status')
+        try:
+            f = file(filename, 'r')
+        except IOError, e:
+            return 'none'
+
+        # fast forward to the end
+        for line in f: pass
+
+        # the first word on the last line is the status
+        return line.split(' ')[0]
+
+    def __str__(self):
+        return self.dir
 
 if __name__ == '__main__':
     rootdir = env.setdefault('ROOTDIR', os.getcwd())
@@ -97,29 +159,27 @@ if __name__ == '__main__':
         workbase = "/tmp/"
 
     workdir = joinpath(workbase, '%s.%s' % (env['USER'], pbs_jobid))
-
-    def echofile(filename, string):
-        try:
-            f = file(joinpath(outdir, filename), 'w')
-            print >>f, string
-            f.flush()
-            f.close()
-        except IOError,e:
-            sys.exit(e)
+    host = socket.gethostname()
 
     os.umask(0022)
 
-    echofile('.start', date())
-    echofile('.pbs_jobid', pbs_jobid)
-    echofile('.pbs_jobname', pbs_jobid)
-    echofile('.host', socket.gethostname())
+    jobdir = JobDir(outdir)
+
+    started = date()
+    jobdir.echofile('.running', started)
+    jobdir.rmfile('.queued')
+    jobdir.echofile('.pbs_jobid', pbs_jobid)
+    jobdir.echofile('.pbs_jobname', pbs_jobid)
+    jobdir.echofile('.host', host)
+
+    jobdir.setstatus('running on %s on %s' % (host, started))
 
     if os.path.isdir(workdir):
         cleandir(workdir)
     else:
         os.mkdir(workdir)
 
-    if os.path.isdir('/z/dist'):
+    if False and os.path.isdir('/z/dist'):
         sync = rsync()
         sync.delete = True
         sync.sudo = True
@@ -130,13 +190,13 @@ if __name__ == '__main__':
     except OSError,e:
         sys.exit(e)
 
-    os.symlink(joinpath(outdir, 'output'), 'status.out')
+    os.symlink(jobdir.file('output'), 'status.out')
 
     args = [ joinpath(basedir, 'm5'), joinpath(basedir, 'run.py') ]
     if not len(args):
         sys.exit("no arguments")
 
-    print 'starting job... %s' % date()
+    print 'starting job... %s' % started
     print ' '.join(args)
     print
     sys.stdout.flush()
@@ -145,7 +205,7 @@ if __name__ == '__main__':
     if not childpid:
         # Execute command
         sys.stdin.close()
-        fd = os.open(joinpath(outdir, "output"),
+        fd = os.open(jobdir.file("output"),
                      os.O_WRONLY | os.O_CREAT | os.O_TRUNC)
         os.dup2(fd, sys.stdout.fileno())
         os.dup2(fd, sys.stderr.fileno())
@@ -170,12 +230,15 @@ if __name__ == '__main__':
             thepid,ec = os.waitpid(childpid, 0)
             if ec:
                 print 'Exit code ', ec
-                echofile('.failure', date())
+                status = 'failure'
             else:
-                echofile('.success', date())
+                status = 'success'
             done = 1
         except OSError:
             pass
 
-    print '\njob complete... %s' % date()
-    echofile('.stop', date())
+    complete = date()
+    print '\njob complete... %s' % complete
+    jobdir.echofile('.%s' % status, complete)
+    jobdir.rmfile('.running')
+    jobdir.setstatus('%s on %s' % (status, complete))