Updates to job scripts to accept more than 15 characters of jobname

Make the Link directory even more useful by working with
sub-directories.

util/pbs/job.py:
    Expose JOBNAME as a separate parameter from PBS_JOBNAME.  If the
    former exists, it is used as the jobname for starting the job, if
    it doesn't exist, PBS_JOBNAME is used.  This is to get around the 15
    character maximum pbs job name length.  While we're at it, shuffle
    things around to hopefully make things a bit more clear.
util/pbs/send.py:
    Make the Link directory functionality more sophisticated, copy
    sub-directories and links to directories.  (we still don't copy
    dotfiles though)
    Add the setname() function to contact pbs and use raj's hack to
    tell the webpage about longer jobnames. (it's gross, don't look)
    truncate the pbs job name to 15 characters so that it works.

--HG--
extra : convert_revision : 4a76b1a1c33721c7ca93e2fbb761f95bc3a2ac69

diff --git a/util/pbs/job.py b/util/pbs/job.py
index 4ead50de9a4c2f66cbad103b8cc08b1b1a145b53..f370862de22c883da5852af9cd51a0dcbcf74c5b 100755 (executable)
--- a/util/pbs/job.py
+++ b/util/pbs/job.py
@@ -83,37 +83,35 @@ def readval(filename):
 
 if __name__ == '__main__':
     rootdir = env.setdefault('ROOTDIR', os.getcwd())
-    jobid = env['PBS_JOBID']
-    jobname = env['PBS_JOBNAME']
-    jobdir = joinpath(rootdir, jobname)
+    pbs_jobid = env['PBS_JOBID']
+    pbs_jobname = env['PBS_JOBNAME']
     basedir = joinpath(rootdir, 'Base')
-    user = env['USER']
-
+    jobname = env.setdefault('JOBNAME', pbs_jobname)
+    jobfile = env.setdefault('JOBFILE', joinpath(basedir, 'test.py'))
+    outdir = env.setdefault('OUTPUT_DIR', joinpath(rootdir, jobname))
     env['POOLJOB'] = 'True'
-    env['OUTPUT_DIR'] = jobdir
-    env['JOBFILE'] = joinpath(basedir, 'test.py')
-    env['JOBNAME'] = jobname
+
+    if os.path.isdir("/work"):
+        workbase = "/work"
+    else:
+        workbase = "/tmp/"
+
+    workdir = joinpath(workbase, '%s.%s' % (env['USER'], pbs_jobid))
 
     def echofile(filename, string):
         try:
-            f = file(joinpath(jobdir, filename), 'w')
+            f = file(joinpath(outdir, filename), 'w')
             print >>f, string
             f.flush()
             f.close()
         except IOError,e:
             sys.exit(e)
 
-    if os.path.isdir("/work"):
-        workbase = "/work"
-    else:
-        workbase = "/tmp/"
-
-    workdir = joinpath(workbase, '%s.%s' % (user, jobid))
-
     os.umask(0022)
 
     echofile('.start', date())
-    echofile('.jobid', jobid)
+    echofile('.pbs_jobid', pbs_jobid)
+    echofile('.pbs_jobname', pbs_jobid)
     echofile('.host', socket.gethostname())
 
     if os.path.isdir(workdir):
@@ -132,7 +130,7 @@ if __name__ == '__main__':
     except OSError,e:
         sys.exit(e)
 
-    os.symlink(joinpath(jobdir, 'output'), 'status.out')
+    os.symlink(joinpath(outdir, 'output'), 'status.out')
 
     args = [ joinpath(basedir, 'm5'), joinpath(basedir, 'run.py') ]
     if not len(args):
@@ -147,7 +145,7 @@ if __name__ == '__main__':
     if not childpid:
         # Execute command
         sys.stdin.close()
-        fd = os.open(joinpath(jobdir, "output"),
+        fd = os.open(joinpath(outdir, "output"),
                      os.O_WRONLY | os.O_CREAT | os.O_TRUNC)
         os.dup2(fd, sys.stdout.fileno())
         os.dup2(fd, sys.stderr.fileno())

diff --git a/util/pbs/send.py b/util/pbs/send.py
index f8ca5209cc00e14b4fbb71a7dab33e11e8cc6d84..ecb0be0ec7c65a4537b9e9c3e805d6abc443e436 100755 (executable)
--- a/util/pbs/send.py
+++ b/util/pbs/send.py
@@ -30,9 +30,9 @@
 
 import os, os.path, re, socket, sys
 from os import environ as env, listdir
-from os.path import basename, isdir, isfile, islink, join as joinpath
+from os.path import basename, isdir, isfile, islink, join as joinpath, normpath
 from filecmp import cmp as filecmp
-from shutil import copyfile
+from shutil import copy
 
 def nfspath(dir):
     if dir.startswith('/.automount/'):
@@ -41,6 +41,38 @@ def nfspath(dir):
         dir = '/n/%s%s' % (socket.gethostname().split('.')[0], dir)
     return dir
 
+def syncdir(srcdir, destdir):
+    srcdir = normpath(srcdir)
+    destdir = normpath(destdir)
+    if not isdir(destdir):
+        sys.exit('destination directory "%s" does not exist' % destdir)
+
+    for root, dirs, files in os.walk(srcdir):
+        root = normpath(root)
+        prefix = os.path.commonprefix([root, srcdir])
+        root = root[len(prefix):]
+        if root.startswith('/'):
+            root = root[1:]
+        for rem in [ d for d in dirs if d.startswith('.') or d == 'SCCS']:
+            dirs.remove(rem)
+
+        for entry in dirs:
+            newdir = joinpath(destdir, root, entry)
+            if not isdir(newdir):
+                os.mkdir(newdir)
+                print 'mkdir', newdir
+
+        for i,d in enumerate(dirs):
+            if islink(joinpath(srcdir, root, d)):
+                dirs[i] = joinpath(d, '.')
+
+        for entry in files:
+            dest = normpath(joinpath(destdir, root, entry))
+            src = normpath(joinpath(srcdir, root, entry))
+            if not isfile(dest) or not filecmp(src, dest):
+                print 'copy %s %s' % (dest, src)
+                copy(src, dest)
+
 progpath = nfspath(sys.path[0])
 progname = basename(sys.argv[0])
 usage = """\
@@ -107,16 +139,7 @@ for arg in args:
 if not listonly and not onlyecho and isdir(linkdir):
     if verbose:
         print 'Checking for outdated files in Link directory'
-    entries = listdir(linkdir)
-    for entry in entries:
-        link = joinpath(linkdir, entry)
-        if not islink(link) or not isfile(link):
-            continue
-
-        base = joinpath(basedir, entry)
-        if not isfile(base) or not filecmp(link, base):
-            print 'Base/%s is different than Link/%s: copying' % (entry, entry)
-            copyfile(link, base)
+    syncdir(linkdir, basedir)
 
 import job, jobfile, pbs
 
@@ -164,6 +187,21 @@ if not onlyecho:
         jl.append(jobname)
     joblist = jl
 
+def setname(jobid, jobname):
+    # since pbs can handle jobnames of 15 characters or less, don't
+    # use the raj hack.
+    if len(jobname) <= 15:
+        return
+
+    import socket
+    s = socket.socket()
+    # Connect to pbs.pool and send the jobid/jobname pair to port
+    # 24465 (Raj didn't realize that there are only 64k ports and
+    # setup inetd to point to port 90001)
+    s.connect(("pbs.pool", 24465))
+    s.send("%s %s\n" % (jobid, jobname))
+    s.close()
+
 for jobname in joblist:
     jobdir = joinpath(rootdir, jobname)
 
@@ -176,10 +214,11 @@ for jobname in joblist:
     qsub = pbs.qsub()
     qsub.pbshost = 'simpool.eecs.umich.edu'
     qsub.stdout = joinpath(jobdir, 'jobout')
-    qsub.name = jobname
+    qsub.name = jobname[:15]
     qsub.join = True
     qsub.node_type = 'FAST'
     qsub.env['ROOTDIR'] = rootdir
+    qsub.env['JOBNAME'] = jobname
     if len(queue):
         qsub.queue = queue
     qsub.build(joinpath(progpath, 'job.py'))
@@ -190,6 +229,8 @@ for jobname in joblist:
     if not onlyecho:
         ec = qsub.do()
         if ec == 0:
-            print 'PBS Jobid:      %s' % qsub.result
+            jobid = qsub.result
+            print 'PBS Jobid:      %s' % jobid
+            setname(jobid, jobname)
         else:
             print 'PBS Failed'