From onionknigh at gmail.com  Thu Apr  1 15:05:46 2010
From: onionknigh at gmail.com (Rickard Westerlund)
Date: Thu, 1 Apr 2010 22:05:46 +0200
Subject: [Swift-user] Problems running example scripts
Message-ID: <g2j12e4d70f1004011305n4d37994byedafdfcfc9e949b9@mail.gmail.com>

Hello,
I've got problems running swift scripts on both Windows and Linux. On
Windows I get errors that indicate my system is expected to be Unix,
output here: http://pastebin.com/CqwFrxDx
Is there something wrong with my configuration or is there a lack of
Windows support?

On Linux I can't get the regexp.swift and foreach.swift to run
properly, they both complain about the character '1' in the regexp
transform. Changing the expression to "\\1count" from "\1count" seems
to make it compile, but then I get the following output:
http://pastebin.com/NDWK4541
Every other example works fine though.

---
Rickard Westerlund


From hategan at mcs.anl.gov  Thu Apr  1 15:16:57 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Thu, 01 Apr 2010 15:16:57 -0500
Subject: [Swift-user] Problems running example scripts
In-Reply-To: <g2j12e4d70f1004011305n4d37994byedafdfcfc9e949b9@mail.gmail.com>
References: <g2j12e4d70f1004011305n4d37994byedafdfcfc9e949b9@mail.gmail.com>
Message-ID: <1270153017.10351.1.camel@localhost>

On Thu, 2010-04-01 at 22:05 +0200, Rickard Westerlund wrote:
> Hello,
> I've got problems running swift scripts on both Windows and Linux. On
> Windows I get errors that indicate my system is expected to be Unix,
> output here: http://pastebin.com/CqwFrxDx
> Is there something wrong with my configuration or is there a lack of
> Windows support?

See this:
http://www.ci.uchicago.edu/swift/guides/userguide.php#tips.windows

> 
> On Linux I can't get the regexp.swift and foreach.swift to run
> properly, they both complain about the character '1' in the regexp
> transform. Changing the expression to "\\1count" from "\1count" seems
> to make it compile, but then I get the following output:
> http://pastebin.com/NDWK4541

That seems to indicate that your tc.data does not have an entry for
"wc". Is that correct?

Mihael




From hategan at mcs.anl.gov  Thu Apr  1 17:03:46 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Thu, 01 Apr 2010 17:03:46 -0500
Subject: [Swift-user] Problems running example scripts
In-Reply-To: <j2x12e4d70f1004011457p784837dfrd9f6ebfb379dc321@mail.gmail.com>
References: <g2j12e4d70f1004011305n4d37994byedafdfcfc9e949b9@mail.gmail.com>
	<1270153017.10351.1.camel@localhost>
	<j2x12e4d70f1004011457p784837dfrd9f6ebfb379dc321@mail.gmail.com>
Message-ID: <1270159426.14641.2.camel@localhost>

On Thu, 2010-04-01 at 23:57 +0200, Rickard Westerlund wrote:
> On Thu, Apr 1, 2010 at 10:16 PM, Mihael Hategan <hategan at mcs.anl.gov> wrote:
> > See this:
> > http://www.ci.uchicago.edu/swift/guides/userguide.php#tips.windows
> That certainly helped, now I'm just getting a "execution failled:
> exception in echo" error and the job failing with an exit code of 252.
> I tried changing paths in tc.data but that didn't help. Removing the
> line for echo gives the "could not find any valid host for task"
> error.

Right. Most of the examples will probably fail to work because the
standard unix executables don't exist in windows (unless you install
cygwin).

So you would need to use some valid windows executables there and write
swift scripts that make sense for them.

> 
> > That seems to indicate that your tc.data does not have an entry for
> > "wc". Is that correct?
> Yep, adding an entry for it solved it.

Ok. Good.




From onionknigh at gmail.com  Thu Apr  1 17:17:46 2010
From: onionknigh at gmail.com (Rickard Westerlund)
Date: Fri, 2 Apr 2010 00:17:46 +0200
Subject: [Swift-user] Problems running example scripts
In-Reply-To: <1270159426.14641.2.camel@localhost>
References: <g2j12e4d70f1004011305n4d37994byedafdfcfc9e949b9@mail.gmail.com> 
	<1270153017.10351.1.camel@localhost>
	<j2x12e4d70f1004011457p784837dfrd9f6ebfb379dc321@mail.gmail.com>
	<1270159426.14641.2.camel@localhost>
Message-ID: <m2y12e4d70f1004011517wd7a5a17dhb61d962ea74dfc20@mail.gmail.com>

On Fri, Apr 2, 2010 at 12:03 AM, Mihael Hategan <hategan at mcs.anl.gov> wrote:
> Right. Most of the examples will probably fail to work because the
> standard unix executables don't exist in windows (unless you install
> cygwin).
I see. I have gnuwin installed and had tried to use the echo from
there, but got the same error. Also tried to use DOS naming to avoid
spaces. Guess I'll have to find a format that works.

Thanks for your fast responses.

---
Rickard Westerlund


From hategan at mcs.anl.gov  Thu Apr  1 17:28:53 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Thu, 01 Apr 2010 17:28:53 -0500
Subject: [Swift-user] Problems running example scripts
In-Reply-To: <m2y12e4d70f1004011517wd7a5a17dhb61d962ea74dfc20@mail.gmail.com>
References: <g2j12e4d70f1004011305n4d37994byedafdfcfc9e949b9@mail.gmail.com>
	<1270153017.10351.1.camel@localhost>
	<j2x12e4d70f1004011457p784837dfrd9f6ebfb379dc321@mail.gmail.com>
	<1270159426.14641.2.camel@localhost>
	<m2y12e4d70f1004011517wd7a5a17dhb61d962ea74dfc20@mail.gmail.com>
Message-ID: <1270160933.16404.0.camel@localhost>

On Fri, 2010-04-02 at 00:17 +0200, Rickard Westerlund wrote:
> On Fri, Apr 2, 2010 at 12:03 AM, Mihael Hategan <hategan at mcs.anl.gov> wrote:
> > Right. Most of the examples will probably fail to work because the
> > standard unix executables don't exist in windows (unless you install
> > cygwin).
> I see. I have gnuwin installed and had tried to use the echo from
> there, but got the same error. Also tried to use DOS naming to avoid
> spaces. Guess I'll have to find a format that works.

Make sure you include the extension (i.e. echo.exe) in tc.data. 




From wilde at mcs.anl.gov  Thu Apr  1 23:44:00 2010
From: wilde at mcs.anl.gov (Michael Wilde)
Date: Thu, 1 Apr 2010 23:44:00 -0500 (CDT)
Subject: [Swift-user] Article on Swift in SciDAC Review
Message-ID: <10961150.1131270183440578.JavaMail.root@zimbra>

This just came out:
  http://www.scidacreview.org/1002/html/swift.html

- Mike

-- 
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory



From amilamad at gmail.com  Fri Apr  2 12:36:51 2010
From: amilamad at gmail.com (Amila Madusanka)
Date: Fri, 2 Apr 2010 23:06:51 +0530
Subject: [Swift-user] Can`t run the first.swift
Message-ID: <u2n30ae5c8f1004021036k4b525926n679730f11e70c677@mail.gmail.com>

when runing  ->  swift first.swift I get the following massage

Execution failed:
        File not found scheduler.xml


please help
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/swift-user/attachments/20100402/2354dd87/attachment.html>

From hategan at mcs.anl.gov  Fri Apr  2 13:03:17 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Fri, 02 Apr 2010 13:03:17 -0500
Subject: [Swift-user] Can`t run the first.swift
In-Reply-To: <u2n30ae5c8f1004021036k4b525926n679730f11e70c677@mail.gmail.com>
References: <u2n30ae5c8f1004021036k4b525926n679730f11e70c677@mail.gmail.com>
Message-ID: <1270231397.2398.1.camel@localhost>

Hi,

You'll need to provide more details than that. Such as what syestem/OS
you're trying to run this on, what you did to download/compile/install
Swift, perhaps the log file that the failed run produced, etc.

Mihael

On Fri, 2010-04-02 at 23:06 +0530, Amila Madusanka wrote:
> when runing  ->  swift first.swift I get the following massage
> 
> 
> Execution failed:
>         File not found scheduler.xml
> 
> 
> 
> 
> please help
> _______________________________________________
> Swift-user mailing list
> Swift-user at ci.uchicago.edu
> http://mail.ci.uchicago.edu/mailman/listinfo/swift-user



From hategan at mcs.anl.gov  Fri Apr  2 13:13:22 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Fri, 02 Apr 2010 13:13:22 -0500
Subject: [Swift-user] Can`t run the first.swift
In-Reply-To: <j2h30ae5c8f1004021108q65fd1a7alebaeae5ccd3c6cf5@mail.gmail.com>
References: <u2n30ae5c8f1004021036k4b525926n679730f11e70c677@mail.gmail.com>
	<1270231397.2398.1.camel@localhost>
	<j2h30ae5c8f1004021108q65fd1a7alebaeae5ccd3c6cf5@mail.gmail.com>
Message-ID: <1270232002.4159.3.camel@localhost>

On Fri, 2010-04-02 at 23:38 +0530, Amila Madusanka wrote:
> I am new to the swift.The OS is Windows Vista Business Edition.I
> extract the swift 0.8 and set the environment variables to the bin.

Which environment variables? Please be specific.

How are you launching swift (i.e. what exact commands are you typing)?

> But for each example it gives   
> 
>  Execution failed:
> >         File not found scheduler.xml
> 

Yes. You already mentioned that.

Please read this:
http://www.chiark.greenend.org.uk/~sgtatham/bugs.html

Mihael



From hategan at mcs.anl.gov  Fri Apr  2 15:01:16 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Fri, 02 Apr 2010 15:01:16 -0500
Subject: [Swift-user] Can`t run the first.swift
In-Reply-To: <p2z30ae5c8f1004021119r64ead8av19ea9838a51764@mail.gmail.com>
References: <u2n30ae5c8f1004021036k4b525926n679730f11e70c677@mail.gmail.com>
	<1270231397.2398.1.camel@localhost>
	<j2h30ae5c8f1004021108q65fd1a7alebaeae5ccd3c6cf5@mail.gmail.com>
	<1270232002.4159.3.camel@localhost>
	<p2z30ae5c8f1004021119r64ead8av19ea9838a51764@mail.gmail.com>
Message-ID: <1270238476.27377.6.camel@localhost>

On Fri, 2010-04-02 at 23:49 +0530, Amila Madusanka wrote:
> I set the "path" environment variable.
> And go to example dir and use command "swift first.swift"   (As in
> Swift Quickstart Guide)

1. You should keep the mailing list CC-ed.

2. We don't support windows much. It's more like we try to make it run
there, but there are no guarantees. So you have two choices:
a) Try to troubleshoot and fix the problems on windows, which would
become part of your GSOC contributions to the project.
b) Install cygwin or use linux

3. I don't think Swift 0.8 works with Windows. I would recommend trying
the SVN version, either the stable branch or trunk. If you are going to
be a GSOC student for Swift, that is the code you will work with.

4. Read
http://www.ci.uchicago.edu/swift/guides/userguide.php#tips.windows

Mihael



From wozniak at mcs.anl.gov  Mon Apr  5 10:11:19 2010
From: wozniak at mcs.anl.gov (Justin M Wozniak)
Date: Mon, 5 Apr 2010 10:11:19 -0500 (CDT)
Subject: [Swift-user] @tostring?
In-Reply-To: <9408326.381681269727849710.JavaMail.root@zimbra>
References: <9408326.381681269727849710.JavaMail.root@zimbra>
Message-ID: <alpine.DEB.2.00.1004051007580.10780@roberts-desktop.mcs.anl.gov>

On Sat, 27 Mar 2010, wilde at mcs.anl.gov wrote:

> (3) Justin has added an @java() primitive to the development trunk; so 
> if there's a Java method among standard or Swift Java classes that you 
> can call to do a needed string operation, that may be useful. I cant 
> recall if that was posted to one of the lists or not.

Sorry for the delay.

Here is the link to the @java note:

http://mail.ci.uchicago.edu/pipermail/swift-user/2010-March/001382.html

Also, I implemented an @tostring() back when I thought @strcat() was 
typechecked.  You can try that if you want.

> === (1) - A "toy" Swift library:
>
> // General Swift Lib Functions
>
> app (file o) echoi (int i)     { echo i stdout=@o;}
> app (file o) echof (float f)   { echo f stdout=@o;}
> app (file o) echob (boolean b) { echo b stdout=@o;}
> app (file o) echos (string s)  { echo s stdout=@o;}
>
> (string s) itostr (int i)
> {
>  file f;
>  f = echoi(i);
>  s = readData(f);
> }
>
> (string s) ftostr (float n)
> {
>  file f;
>  f = echof(n);
>  s = readData(f);
> }
>
> (int n) strtoi (string s)
> {
>  file f;
>  f = echos(s);
>  n = readData(f);
> }
>
> (float n) strtof (string s)
> {
>  file f;
>  f = echos(s);
>  n = readData(f);
> }
>
> app (file o) sprintfsApp (string fmt, string e[])
> {
>  sprintfs fmt e stdout=@o;
> }
>
> (string s) sprintfs (string fmt, string e[])
> {
>  file f;
>  f = sprintfsApp(fmt,e);
>  s = readData(f);
> }
>
> === (2) swiftshell:
>
> login1$ more shelldemo.swift swiftshell
> ::::::::::::::
> shelldemo.swift
> ::::::::::::::
> type file;
>
> app (file o) cat (file i)
> {
>  shell " ( cat " @i "; date; hostname ) | grep . " stdout=@o;
> }
>
> file data<"data.txt">;
> file out<"out.txt">;
> out = cat(data);
> ::::::::::::::
> swiftshell
> ::::::::::::::
> bash -c "$*"
> login1$ grep shell tc
> localhost 	shell	/home/wilde/swift/lab/swiftshell	null	null	null
> mcs	 	shell	/home/wilde/swift/lab/swiftshell	null	null	null
> login1$
>
> === (3) @java()
>
> Justin previously posted this to the list:
>
> "         If you can check out the latest Swift from trunk I've added some
> features that might help you out here.  There's a new built-in function @java()
> that allows you to call into an existing Java library.  You can call into the
> Java Platform or into your CLASSPATH.
>         Here is one example:
>
> (float result) sin(float x) {
>   result = @java("java.lang.Math", "sin", x);
> }
>
> float x = 0.5;
> float y = sin(x);
>
> trace("sin", x, y);
>
> Note that you currently have to assign the result of @java() to a variable."
>
>
>
>
> ----- "Andriy Fedorov" <fedorov at bwh.harvard.edu> wrote:
>
>> On Sat, Mar 27, 2010 at 17:34, Michael Wilde <wilde at mcs.anl.gov>
>> wrote:
>>> Andriy, Im pretty sure @strcat() will take an int and return a
>> string.
>>>
>>
>> Mike, yes, you are right -- there was another error in my script,
>> @strcat indeed works with int.
>>
>> One more basic question (sorry if I missed this in the guide): is
>> there a way to get the size of an array? I would like to split a
>> string, and get the last item in the array it returns. Is this
>> possible?
>>
>>
>>> - Mike
>>>
>>> ----- "Andriy Fedorov" <fedorov at bwh.harvard.edu> wrote:
>>>
>>>> Hi,
>>>>
>>>> Is it possible to convert int type to string type in Swift?
>>>>
>>>> I see @toint, but not @tostring, and it looks like I am not able
>> to
>>>> pass int to @strcat().
>>>>
>>>> Thanks
>>>>
>>>> --
>>>> Andriy Fedorov, Ph.D.
>>>>
>>>> Research Fellow
>>>> Brigham and Women's Hospital
>>>> Harvard Medical School
>>>> 75 Francis Street
>>>> Boston, MA 02115 USA
>>>> fedorov at bwh.harvard.edu
>>>> _______________________________________________
>>>> Swift-user mailing list
>>>> Swift-user at ci.uchicago.edu
>>>> http://mail.ci.uchicago.edu/mailman/listinfo/swift-user
>>>
>>> --
>>> Michael Wilde
>>> Computation Institute, University of Chicago
>>> Mathematics and Computer Science Division
>>> Argonne National Laboratory
>>>
>>>
>
>

-- 
Justin M Wozniak


From marcin at galton.uchicago.edu  Tue Apr  6 00:20:50 2010
From: marcin at galton.uchicago.edu (Marcin Hitczenko)
Date: Tue, 6 Apr 2010 00:20:50 -0500 (CDT)
Subject: [Swift-user] swift and fusion
Message-ID: <40547.207.181.247.181.1270531250.squirrel@galton.uchicago.edu>

Hi,

I am using swift to submit several R jobs on fusion and am trying to
determine whether or not I am making use of all the available cores on
each node (I believe there are 8 cores on each node). I submitted 10
really short jobs to try to see if I could determine what was going on,
but I don't really know what to look for.

In case it is useful, I am attaching the .log file, sites.xml, and an info
file for one of the jobs.

Thanks for your help,

Marcin
-------------- next part --------------
A non-text attachment was scrubbed...
Name: sites.xml
Type: text/xml
Size: 375 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/swift-user/attachments/20100406/cfe8f2ff/attachment.xml>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: environment_setup-20100405-2320-oi51qh72.log
Type: application/octet-stream
Size: 341881 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/swift-user/attachments/20100406/cfe8f2ff/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: RBATCH-chvmf5qj-info
Type: application/octet-stream
Size: 2128 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/swift-user/attachments/20100406/cfe8f2ff/attachment-0001.obj>

From aespinosa at cs.uchicago.edu  Sat Apr 10 00:30:24 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Sat, 10 Apr 2010 00:30:24 -0500
Subject: [Swift-user] throttling mapping threads
Message-ID: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>

Hi,

How do you throttle the number of mapping threads like
foreach.max.threads ?  I have this workflow whose mapper accesses a
mysql database that makes a 400+ access at a time.

I can change foreach.max.threads, but that then would affect the
number of jobs I could send at a time too.

-- 
Allan M. Espinosa <http://amespinosa.wordpress.com>
PhD student, Computer Science
University of Chicago <http://people.cs.uchicago.edu/~aespinosa>


From hategan at mcs.anl.gov  Sat Apr 10 00:36:16 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Sat, 10 Apr 2010 00:36:16 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
Message-ID: <1270877776.14163.1.camel@localhost>

On Sat, 2010-04-10 at 00:30 -0500, Allan Espinosa wrote:
> Hi,
> 
> How do you throttle the number of mapping threads like
> foreach.max.threads ?  I have this workflow whose mapper accesses a
> mysql database that makes a 400+ access at a time.

There is no throttling for those unfortunately.

Any way you can bulk the mapping invocations?




From aespinosa at cs.uchicago.edu  Sat Apr 10 00:45:28 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Sat, 10 Apr 2010 00:45:28 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <1270877776.14163.1.camel@localhost>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
Message-ID: <s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>

Oh well...

Yeah, I'm actually working on caching the queries somewhere (via memcache)

Thanks!
-Allan

2010/4/10 Mihael Hategan <hategan at mcs.anl.gov>:
> On Sat, 2010-04-10 at 00:30 -0500, Allan Espinosa wrote:
>> Hi,
>>
>> How do you throttle the number of mapping threads like
>> foreach.max.threads ? ?I have this workflow whose mapper accesses a
>> mysql database that makes a 400+ access at a time.
>
> There is no throttling for those unfortunately.
>
> Any way you can bulk the mapping invocations?
>
>
>


From aespinosa at cs.uchicago.edu  Mon Apr 12 16:11:01 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Mon, 12 Apr 2010 16:11:01 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
Message-ID: <1271106661.2089.8.camel@origin>

I have tried caching queries and still had problems.  I just made a C
version of my mapper because i thought invoking the interpreter might be
too much; still had problems. 

I guess i'll now limit the foreach.max.threads then...
-Allan

On Sab, 2010-04-10 at 00:45 -0500, Allan Espinosa wrote:
> Oh well...
> 
> Yeah, I'm actually working on caching the queries somewhere (via memcache)
> 
> Thanks!
> -Allan
> 
> 2010/4/10 Mihael Hategan <hategan at mcs.anl.gov>:
> > On Sat, 2010-04-10 at 00:30 -0500, Allan Espinosa wrote:
> >> Hi,
> >>
> >> How do you throttle the number of mapping threads like
> >> foreach.max.threads ?  I have this workflow whose mapper accesses a
> >> mysql database that makes a 400+ access at a time.
> >
> > There is no throttling for those unfortunately.
> >
> > Any way you can bulk the mapping invocations?
> >
> >
> >




From aespinosa at cs.uchicago.edu  Mon Apr 12 19:28:17 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Mon, 12 Apr 2010 19:28:17 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <1271106661.2089.8.camel@origin>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin>
Message-ID: <1271118497.2089.10.camel@origin>

Aha!  I just delegated my mapper to an app function + readdata
incantation.  Hopefully this will work

-Allan

On Lun, 2010-04-12 at 16:11 -0500, Allan Espinosa wrote:
> I have tried caching queries and still had problems.  I just made a C
> version of my mapper because i thought invoking the interpreter might be
> too much; still had problems. 
> 
> I guess i'll now limit the foreach.max.threads then...
> -Allan
> 
> On Sab, 2010-04-10 at 00:45 -0500, Allan Espinosa wrote:
> > Oh well...
> > 
> > Yeah, I'm actually working on caching the queries somewhere (via memcache)
> > 
> > Thanks!
> > -Allan
> > 
> > 2010/4/10 Mihael Hategan <hategan at mcs.anl.gov>:
> > > On Sat, 2010-04-10 at 00:30 -0500, Allan Espinosa wrote:
> > >> Hi,
> > >>
> > >> How do you throttle the number of mapping threads like
> > >> foreach.max.threads ?  I have this workflow whose mapper accesses a
> > >> mysql database that makes a 400+ access at a time.
> > >
> > > There is no throttling for those unfortunately.
> > >
> > > Any way you can bulk the mapping invocations?



From benc at hawaga.org.uk  Tue Apr 13 09:02:58 2010
From: benc at hawaga.org.uk (Ben Clifford)
Date: Tue, 13 Apr 2010 14:02:58 +0000 (GMT)
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <1271118497.2089.10.camel@origin>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin> <1271118497.2089.10.camel@origin>
Message-ID: <Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>


> Aha!  I just delegated my mapper to an app function + readdata
> incantation.  Hopefully this will work

one invocation per mapping or one for the whole script?

-- 



From aespinosa at cs.uchicago.edu  Tue Apr 13 09:04:57 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Tue, 13 Apr 2010 09:04:57 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin> <1271118497.2089.10.camel@origin>
	<Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>
Message-ID: <l2v50b07b4b1004130704g82187456ua667bdc4db1fffa4@mail.gmail.com>

one for the whole script which is invoked a lot of times

2010/4/13 Ben Clifford <benc at hawaga.org.uk>:
>
>> Aha! ?I just delegated my mapper to an app function + readdata
>> incantation. ?Hopefully this will work
>
> one invocation per mapping or one for the whole script?
>
> --
>
>
>


From benc at hawaga.org.uk  Tue Apr 13 11:11:12 2010
From: benc at hawaga.org.uk (Ben Clifford)
Date: Tue, 13 Apr 2010 16:11:12 +0000 (GMT)
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <l2v50b07b4b1004130704g82187456ua667bdc4db1fffa4@mail.gmail.com>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin> <1271118497.2089.10.camel@origin> 
	<Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>
	<l2v50b07b4b1004130704g82187456ua667bdc4db1fffa4@mail.gmail.com>
Message-ID: <Pine.LNX.4.64.1004131610590.25367@dildano.hawaga.org.uk>

> one for the whole script which is invoked a lot of times


you're invoking the same swiftscript a whole lot of times?

-- 



From aespinosa at cs.uchicago.edu  Tue Apr 13 11:52:52 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Tue, 13 Apr 2010 11:52:52 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <Pine.LNX.4.64.1004131610590.25367@dildano.hawaga.org.uk>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin> <1271118497.2089.10.camel@origin>
	<Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>
	<l2v50b07b4b1004130704g82187456ua667bdc4db1fffa4@mail.gmail.com>
	<Pine.LNX.4.64.1004131610590.25367@dildano.hawaga.org.uk>
Message-ID: <r2m50b07b4b1004130952y465e6597g47905c65f2331740@mail.gmail.com>

oh i was referring to my ext mapper script
1 swift script invoked once, inside of it is a foreach doing 400+ ext mappings.

-Allan

2010/4/13 Ben Clifford <benc at hawaga.org.uk>:
>> one for the whole script which is invoked a lot of times
>
>
> you're invoking the same swiftscript a whole lot of times?
>
> --
>
>
>



-- 
Allan M. Espinosa <http://amespinosa.wordpress.com>
PhD student, Computer Science
University of Chicago <http://people.cs.uchicago.edu/~aespinosa>


From hategan at mcs.anl.gov  Wed Apr 14 00:45:42 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Wed, 14 Apr 2010 00:45:42 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <r2m50b07b4b1004130952y465e6597g47905c65f2331740@mail.gmail.com>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin> <1271118497.2089.10.camel@origin>
	<Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>
	<l2v50b07b4b1004130704g82187456ua667bdc4db1fffa4@mail.gmail.com>
	<Pine.LNX.4.64.1004131610590.25367@dildano.hawaga.org.uk>
	<r2m50b07b4b1004130952y465e6597g47905c65f2331740@mail.gmail.com>
Message-ID: <1271223942.7014.8.camel@localhost>

So I looked at the problem in a bit more detail.

The part that starts the external process is run from the karajan worker
threads which are limited in number (somewhere between 1 and 8)*. So
that's the maximum number of concurrent invocations of an external
mapper. Do you actually see all 400 of them run at once?

Mihael

(*) Funny thing that for cooperative multi-tasking such non-cooperating
tasks as the example above were the thing that prompted the development
of preemptive multi-tasking. And yet here's an example where, due to
lazy coding, it turns out to be helpful.

On Tue, 2010-04-13 at 11:52 -0500, Allan Espinosa wrote:
> oh i was referring to my ext mapper script
> 1 swift script invoked once, inside of it is a foreach doing 400+ ext mappings.
> 
> -Allan
> 
> 2010/4/13 Ben Clifford <benc at hawaga.org.uk>:
> >> one for the whole script which is invoked a lot of times
> >
> >
> > you're invoking the same swiftscript a whole lot of times?
> >
> > --
> >
> >
> >
> 
> 
> 



From aespinosa at cs.uchicago.edu  Wed Apr 14 02:41:07 2010
From: aespinosa at cs.uchicago.edu (Allan Espinosa)
Date: Wed, 14 Apr 2010 02:41:07 -0500
Subject: [Swift-user] throttling mapping threads
In-Reply-To: <1271223942.7014.8.camel@localhost>
References: <i2g50b07b4b1004092230rf7f3ef2s79c8a26c9d3212a0@mail.gmail.com>
	<1270877776.14163.1.camel@localhost>
	<s2t50b07b4b1004092245peb3d1ab1h4a1daf7b3700589a@mail.gmail.com>
	<1271106661.2089.8.camel@origin> <1271118497.2089.10.camel@origin>
	<Pine.LNX.4.64.1004131402320.7388@dildano.hawaga.org.uk>
	<l2v50b07b4b1004130704g82187456ua667bdc4db1fffa4@mail.gmail.com>
	<Pine.LNX.4.64.1004131610590.25367@dildano.hawaga.org.uk>
	<r2m50b07b4b1004130952y465e6597g47905c65f2331740@mail.gmail.com>
	<1271223942.7014.8.camel@localhost>
Message-ID: <l2v50b07b4b1004140041v55f37d76s17896decaafba015@mail.gmail.com>

Moving the thread to swift-devel.

Ah you're right.  i did a thread trace in my old setup:

$ jstack 8026 | grep "state =" | cat -n
     1  Thread 8051: (state = BLOCKED)
     2  Thread 8050: (state = BLOCKED)
     3  Thread 8047: (state = BLOCKED)
     4  Thread 8045: (state = BLOCKED)
     5  Thread 8043: (state = BLOCKED)
     6  Thread 8042: (state = BLOCKED)
     7  Thread 8041: (state = BLOCKED)
     8  Thread 8040: (state = BLOCKED)
     9  Thread 8034: (state = BLOCKED)
    10  Thread 8033: (state = BLOCKED)
    11  Thread 8032: (state = BLOCKED)
    12  Thread 8026: (state = BLOCKED)

Also I added a line in my ext mapper that creates a files.  after 13
minutes from the time i started the workflow, only 12 files were
created.  i'll check with the new version's swift.log on how long
these files are expected to come.

-Alla

2010/4/14 Mihael Hategan <hategan at mcs.anl.gov>:
> So I looked at the problem in a bit more detail.
>
> The part that starts the external process is run from the karajan worker
> threads which are limited in number (somewhere between 1 and 8)*. So
> that's the maximum number of concurrent invocations of an external
> mapper. Do you actually see all 400 of them run at once?
>
> Mihael
>
> (*) Funny thing that for cooperative multi-tasking such non-cooperating
> tasks as the example above were the thing that prompted the development
> of preemptive multi-tasking. And yet here's an example where, due to
> lazy coding, it turns out to be helpful.
>
> On Tue, 2010-04-13 at 11:52 -0500, Allan Espinosa wrote:
>> oh i was referring to my ext mapper script
>> 1 swift script invoked once, inside of it is a foreach doing 400+ ext mappings.
>>
>> -Allan
>>
>> 2010/4/13 Ben Clifford <benc at hawaga.org.uk>:
>> >> one for the whole script which is invoked a lot of times
>> >
>> >
>> > you're invoking the same swiftscript a whole lot of times?


From tstitt at cscs.ch  Thu Apr 15 08:06:57 2010
From: tstitt at cscs.ch (Stitt  Timothy)
Date: Thu, 15 Apr 2010 15:06:57 +0200
Subject: [Swift-user] Can't Access Swift Quickstart and Tutorial HTML Pages
Message-ID: <D003094E-8E5B-406F-B6D4-471E87508C67@cscs.ch>

Dear Swift Development Team,

I was interested in playing around with Swift for the first time today but I can't seem to access the Quickstart and Tutorial HTML pages on the project website. I am receiving some PHP related warnings and errors when I access the links.

I would be grateful if you could look into this...particularly for the Quickstart guide that doesn't have an accompanying PDF version.

Regards,

Tim.

-----------------------------

Timothy Stitt Ph.D
National Support Services
Swiss National Supercomputing Centre (CSCS), Switzerland
Phone: +41 (0) 91 610 8233

Email: stitt(at)cscs.ch<http://cscs.ch>



From wilde at mcs.anl.gov  Thu Apr 15 10:26:51 2010
From: wilde at mcs.anl.gov (Michael Wilde)
Date: Thu, 15 Apr 2010 10:26:51 -0500 (CDT)
Subject: [Swift-user] Can't Access Swift Quickstart and Tutorial HTML Pages
In-Reply-To: <D003094E-8E5B-406F-B6D4-471E87508C67@cscs.ch>
Message-ID: <12535219.366391271345211229.JavaMail.root@zimbra>

Thanks for spotting this, Tim. Ive notified our sysadmin team to correct the problem (caused by web server re-orgs done yesterday). I hope this will be fixed by the end of the day (CDT) if not sooner.

- Mike

----- "Stitt Timothy" <tstitt at cscs.ch> wrote:

> Dear Swift Development Team,
> 
> I was interested in playing around with Swift for the first time today
> but I can't seem to access the Quickstart and Tutorial HTML pages on
> the project website. I am receiving some PHP related warnings and
> errors when I access the links.
> 
> I would be grateful if you could look into this...particularly for the
> Quickstart guide that doesn't have an accompanying PDF version.
> 
> Regards,
> 
> Tim.
> 
> -----------------------------
> 
> Timothy Stitt Ph.D
> National Support Services
> Swiss National Supercomputing Centre (CSCS), Switzerland
> Phone: +41 (0) 91 610 8233
> 
> Email: stitt(at)cscs.ch<http://cscs.ch>
> 
> _______________________________________________
> Swift-user mailing list
> Swift-user at ci.uchicago.edu
> http://mail.ci.uchicago.edu/mailman/listinfo/swift-user

-- 
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory



From wilde at mcs.anl.gov  Sat Apr 17 14:54:16 2010
From: wilde at mcs.anl.gov (Michael Wilde)
Date: Sat, 17 Apr 2010 14:54:16 -0500 (CDT)
Subject: [Swift-user] Swift documentation is back online
In-Reply-To: <12535219.366391271345211229.JavaMail.root@zimbra>
Message-ID: <27037321.419261271534056463.JavaMail.root@zimbra>

Timothy, all - my apologies for the outage.

- Mike

----- "Michael Wilde" <wilde at mcs.anl.gov> wrote:

> Thanks for spotting this, Tim. Ive notified our sysadmin team to
> correct the problem (caused by web server re-orgs done yesterday). I
> hope this will be fixed by the end of the day (CDT) if not sooner.
> 
> - Mike
> 
> ----- "Stitt Timothy" <tstitt at cscs.ch> wrote:
> 
> > Dear Swift Development Team,
> >
> > I was interested in playing around with Swift for the first time
> today
> > but I can't seem to access the Quickstart and Tutorial HTML pages on
> > the project website. I am receiving some PHP related warnings and
> > errors when I access the links.
> >
> > I would be grateful if you could look into this...particularly for
> the
> > Quickstart guide that doesn't have an accompanying PDF version.
> >
> > Regards,
> >
> > Tim.
> >
> > -----------------------------
> >
> > Timothy Stitt Ph.D
> > National Support Services
> > Swiss National Supercomputing Centre (CSCS), Switzerland
> > Phone: +41 (0) 91 610 8233
> >
> > Email: stitt(at)cscs.ch<http://cscs.ch>
> >
> > _______________________________________________
> > Swift-user mailing list
> > Swift-user at ci.uchicago.edu
> > http://mail.ci.uchicago.edu/mailman/listinfo/swift-user
> 
> --
> Michael Wilde
> Computation Institute, University of Chicago
> Mathematics and Computer Science Division
> Argonne National Laboratory

-- 
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory



From tstitt at cscs.ch  Mon Apr 19 04:32:07 2010
From: tstitt at cscs.ch (Stitt  Timothy)
Date: Mon, 19 Apr 2010 11:32:07 +0200
Subject: [Swift-user] Re: Swift documentation is back online
In-Reply-To: <27037321.419261271534056463.JavaMail.root@zimbra>
References: <27037321.419261271534056463.JavaMail.root@zimbra>
Message-ID: <432ED015-BEBE-41A5-90A4-9D9B81D29C4C@cscs.ch>

Now worries Mike...thanks for that. All working now.

Cheers,

Tim.

On Apr 17, 2010, at 9:54 PM, Michael Wilde wrote:

Timothy, all - my apologies for the outage.

- Mike

----- "Michael Wilde" <wilde at mcs.anl.gov<mailto:wilde at mcs.anl.gov>> wrote:

Thanks for spotting this, Tim. Ive notified our sysadmin team to
correct the problem (caused by web server re-orgs done yesterday). I
hope this will be fixed by the end of the day (CDT) if not sooner.

- Mike

----- "Stitt Timothy" <tstitt at cscs.ch<mailto:tstitt at cscs.ch>> wrote:

Dear Swift Development Team,

I was interested in playing around with Swift for the first time
today
but I can't seem to access the Quickstart and Tutorial HTML pages on
the project website. I am receiving some PHP related warnings and
errors when I access the links.

I would be grateful if you could look into this...particularly for
the
Quickstart guide that doesn't have an accompanying PDF version.

Regards,

Tim.

-----------------------------

Timothy Stitt Ph.D
National Support Services
Swiss National Supercomputing Centre (CSCS), Switzerland
Phone: +41 (0) 91 610 8233

Email: stitt(at)cscs.ch<http://cscs.ch>

_______________________________________________
Swift-user mailing list
Swift-user at ci.uchicago.edu<mailto:Swift-user at ci.uchicago.edu>
http://mail.ci.uchicago.edu/mailman/listinfo/swift-user

--
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory

--
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory


-----------------------------

Timothy Stitt Ph.D
National Support Services
Swiss National Supercomputing Centre (CSCS), Switzerland
Phone: +41 (0) 91 610 8233

Email: stitt(at)cscs.ch<http://cscs.ch>



From wilde at mcs.anl.gov  Tue Apr 27 06:04:48 2010
From: wilde at mcs.anl.gov (wilde at mcs.anl.gov)
Date: Tue, 27 Apr 2010 06:04:48 -0500 (CDT)
Subject: [Swift-user] Set maxtime > maxwalltime or your script will hang
In-Reply-To: <8065554.659051272366271851.JavaMail.root@zimbra>
Message-ID: <11830128.659071272366288457.JavaMail.root@zimbra>

[cc'ing swift-user]

Marcin, 

Quick answer: Since you changed the maxwalltime in tc.data to 10 minutes, change the "maxtime" setting in sites.xml to N times 10 minutes *plus* 1 minute. The "plus" is important.

Long answer:

Swift deducts a small fraction of maxtime to use to cleanly shut down the PBS job. The default for this "reserve time" (see the Users Guide) is 10 seconds. So while before it was happily fitting 15-second jobs (the prior setting you had for maxwalltime) into (600-10) second slots, now (with maxwalltime increased to 10 minutes) it could not find any slots into which it could fit a 10 *minute* job. Unfortunately, at the moment, Swift just hangs, continuing to try to find a slot until the maxtime time runs out and the PBS jobs shut down. (There are "good" reasons for this "bad" behavior, which we need to fix)

So bottom line: make maxtime some multiple of maxwalltime, and add a bit to maxtime (say 1 minute, or at least 10 seconds). Note that since maxwalltime is just an *estimate* of how long you expect the apptask to run for, this division is by necessity approximate. After an apptask finishes on a coaster worker CPU, the CPU becomes free, and has some varying amount of time left before the coaster worker expires. Then the process repeats, and Swift again uses maxwalltime to see if there is a coaster worker with at least maxwalltime remaining that can run the next apptask.

- Mike

ps. How about if from now on, if you forget to cc swift-user on these questions, then I will just cc the list on my replies. This has 2 benefits: other users benefit from your questions and my answers, and other swift developers and users can contribute more advice, suggest betters approach, or correct me when I goof.

You should join the swift-user list if you have not already done so:
  http://www.ci.uchicago.edu/swift/support/index.php

Thanks!

----- "Marcin Hitczenko" <marcin at galton.uchicago.edu> wrote:

> Hi Mike,
> 
> Sorry to be such a pain, but I can't get the jobs to run again. I am
> still
> running environment_setup.swift and have not made any changes to the
> pbscoast.xml file other than to change walltime. I am using the same
> tc.data file. I again run into the problem where it seems to run for
> a
> minute, then fail and restart and it does so over and over. I logged
> into
> the node and it didn't seem to be running my stuff. Also, the output
> files
> are not being written.
> 
> I am not sure why it would not work now, though I ran it and it
> worked
> before.
> 
> Marcin
> 
> 
> 



From marcin at galton.uchicago.edu  Tue Apr 27 10:56:17 2010
From: marcin at galton.uchicago.edu (Marcin Hitczenko)
Date: Tue, 27 Apr 2010 10:56:17 -0500 (CDT)
Subject: [Swift-user] Re: Set maxtime > maxwalltime or your script will hang
Message-ID: <46971.207.181.247.181.1272383777.squirrel@galton.uchicago.edu>

Hi Mike,

Two unrelated questions:

1. It seems then if I have 8 apptasks calls each of which is one hour
(maxwalltime is 01:00:00). I should be able to accomplish this by
submitting one PBS job for one node/ 8 cores and having each core run a
different task. If I understand correctly, this should take on the order
of 1 hour plus some extra. But, we would set maxwalltime to 60*60*8, or 8
hours. Why should it run for 8 hours? I feel as I may still be
misunderstanding the distinctions.

2. I submitted a job and would like to cancel it (because it has an
error). I use qdel to cancel the job, but within a few minutes a new job
restarts. I take it this is swift retrying. How do I actually cancel the
entire job? I have done qdel quite a few times, but jobs keep popping up?

Thanks,

Marcin


> [cc'ing swift-user]
>
> Marcin,
>
> Quick answer: Since you changed the maxwalltime in tc.data to 10 minutes,
> change the "maxtime" setting in sites.xml to N times 10 minutes *plus* 1
> minute. The "plus" is important.
>
> Long answer:
>
> Swift deducts a small fraction of maxtime to use to cleanly shut down the
> PBS job. The default for this "reserve time" (see the Users Guide) is 10
> seconds. So while before it was happily fitting 15-second jobs (the prior
> setting you had for maxwalltime) into (600-10) second slots, now (with
> maxwalltime increased to 10 minutes) it could not find any slots into
> which it could fit a 10 *minute* job. Unfortunately, at the moment, Swift
> just hangs, continuing to try to find a slot until the maxtime time runs
> out and the PBS jobs shut down. (There are "good" reasons for this "bad"
> behavior, which we need to fix)
>
> So bottom line: make maxtime some multiple of maxwalltime, and add a bit
> to maxtime (say 1 minute, or at least 10 seconds). Note that since
> maxwalltime is just an *estimate* of how long you expect the apptask to
> run for, this division is by necessity approximate. After an apptask
> finishes on a coaster worker CPU, the CPU becomes free, and has some
> varying amount of time left before the coaster worker expires. Then the
> process repeats, and Swift again uses maxwalltime to see if there is a
> coaster worker with at least maxwalltime remaining that can run the next
> apptask.
>
> - Mike
>
> ps. How about if from now on, if you forget to cc swift-user on these
> questions, then I will just cc the list on my replies. This has 2
> benefits: other users benefit from your questions and my answers, and
> other swift developers and users can contribute more advice, suggest
> betters approach, or correct me when I goof.
>
> You should join the swift-user list if you have not already done so:
>   http://www.ci.uchicago.edu/swift/support/index.php
>
> Thanks!
>
> ----- "Marcin Hitczenko" <marcin at galton.uchicago.edu> wrote:
>
>> Hi Mike,
>>
>> Sorry to be such a pain, but I can't get the jobs to run again. I am
>> still
>> running environment_setup.swift and have not made any changes to the
>> pbscoast.xml file other than to change walltime. I am using the same
>> tc.data file. I again run into the problem where it seems to run for
>> a
>> minute, then fail and restart and it does so over and over. I logged
>> into
>> the node and it didn't seem to be running my stuff. Also, the output
>> files
>> are not being written.
>>
>> I am not sure why it would not work now, though I ran it and it
>> worked
>> before.
>>
>> Marcin
>>
>>
>>
>



From wilde at mcs.anl.gov  Tue Apr 27 11:06:09 2010
From: wilde at mcs.anl.gov (Michael Wilde)
Date: Tue, 27 Apr 2010 11:06:09 -0500 (CDT)
Subject: [Swift-user] Re: Set maxtime > maxwalltime or your script will hang
In-Reply-To: <46971.207.181.247.181.1272383777.squirrel@galton.uchicago.edu>
Message-ID: <17326438.669351272384369740.JavaMail.root@zimbra>

Marcin,

----- "Marcin Hitczenko" <marcin at galton.uchicago.edu> wrote:

> Hi Mike,
> 
> Two unrelated questions:
> 
> 1. It seems then if I have 8 apptasks calls each of which is one hour
> (maxwalltime is 01:00:00). I should be able to accomplish this by
> submitting one PBS job for one node/ 8 cores and having each core run
> a
> different task. If I understand correctly, this should take on the
> order
> of 1 hour plus some extra. But, we would set maxwalltime to 60*60*8,
> or 8
> hours. Why should it run for 8 hours? I feel as I may still be
> misunderstanding the distinctions.

For this case, set maxwalltime to 01:00:00 (ie the estimated duration of each app() task) and maxtime to 3700 (1 hour with 100 secs "reserve")

One caveat, though, for running on the Fusion cluster: I know that if you ask for the batch queue, it insists that you request at least 2 nodes; if you ask for one, your job is rejected. We need to test exactly how to configure coasters for 1 node on that system. You can try removing the queue element from your sites.xml, and see how it behaves. I need to do more testing on that system and post the results.

> 
> 2. I submitted a job and would like to cancel it (because it has an
> error). I use qdel to cancel the job, but within a few minutes a new
> job
> restarts. I take it this is swift retrying. How do I actually cancel
> the
> entire job? I have done qdel quite a few times, but jobs keep popping
> up?

I think the best way to clean up is to interrupt/kill the swift command with a ^C,
and then it should clean up its PBS jobs. I think this has been working for me; let us know if that doesnt work for you.

- Mike


> Thanks,
> 
> Marcin
> 
> 
> > [cc'ing swift-user]
> >
> > Marcin,
> >
> > Quick answer: Since you changed the maxwalltime in tc.data to 10
> minutes,
> > change the "maxtime" setting in sites.xml to N times 10 minutes
> *plus* 1
> > minute. The "plus" is important.
> >
> > Long answer:
> >
> > Swift deducts a small fraction of maxtime to use to cleanly shut
> down the
> > PBS job. The default for this "reserve time" (see the Users Guide)
> is 10
> > seconds. So while before it was happily fitting 15-second jobs (the
> prior
> > setting you had for maxwalltime) into (600-10) second slots, now
> (with
> > maxwalltime increased to 10 minutes) it could not find any slots
> into
> > which it could fit a 10 *minute* job. Unfortunately, at the moment,
> Swift
> > just hangs, continuing to try to find a slot until the maxtime time
> runs
> > out and the PBS jobs shut down. (There are "good" reasons for this
> "bad"
> > behavior, which we need to fix)
> >
> > So bottom line: make maxtime some multiple of maxwalltime, and add a
> bit
> > to maxtime (say 1 minute, or at least 10 seconds). Note that since
> > maxwalltime is just an *estimate* of how long you expect the apptask
> to
> > run for, this division is by necessity approximate. After an
> apptask
> > finishes on a coaster worker CPU, the CPU becomes free, and has
> some
> > varying amount of time left before the coaster worker expires. Then
> the
> > process repeats, and Swift again uses maxwalltime to see if there is
> a
> > coaster worker with at least maxwalltime remaining that can run the
> next
> > apptask.
> >
> > - Mike
> >
> > ps. How about if from now on, if you forget to cc swift-user on
> these
> > questions, then I will just cc the list on my replies. This has 2
> > benefits: other users benefit from your questions and my answers,
> and
> > other swift developers and users can contribute more advice,
> suggest
> > betters approach, or correct me when I goof.
> >
> > You should join the swift-user list if you have not already done
> so:
> >   http://www.ci.uchicago.edu/swift/support/index.php
> >
> > Thanks!
> >
> > ----- "Marcin Hitczenko" <marcin at galton.uchicago.edu> wrote:
> >
> >> Hi Mike,
> >>
> >> Sorry to be such a pain, but I can't get the jobs to run again. I
> am
> >> still
> >> running environment_setup.swift and have not made any changes to
> the
> >> pbscoast.xml file other than to change walltime. I am using the
> same
> >> tc.data file. I again run into the problem where it seems to run
> for
> >> a
> >> minute, then fail and restart and it does so over and over. I
> logged
> >> into
> >> the node and it didn't seem to be running my stuff. Also, the
> output
> >> files
> >> are not being written.
> >>
> >> I am not sure why it would not work now, though I ran it and it
> >> worked
> >> before.
> >>
> >> Marcin
> >>
> >>
> >>
> >

-- 
Michael Wilde
Computation Institute, University of Chicago
Mathematics and Computer Science Division
Argonne National Laboratory



From hategan at mcs.anl.gov  Tue Apr 27 11:48:35 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Tue, 27 Apr 2010 11:48:35 -0500
Subject: [Swift-user] Set maxtime > maxwalltime or your script will hang
In-Reply-To: <11830128.659071272366288457.JavaMail.root@zimbra>
References: <11830128.659071272366288457.JavaMail.root@zimbra>
Message-ID: <1272386915.3670.2.camel@localhost>

On Tue, 2010-04-27 at 06:04 -0500, wilde at mcs.anl.gov wrote:
> [cc'ing swift-user]
> 
> Marcin, 
> 
> Quick answer: Since you changed the maxwalltime in tc.data to 10
> minutes, change the "maxtime" setting in sites.xml to N times 10
> minutes *plus* 1 minute. The "plus" is important.
> 
> Long answer:
> 
> Swift deducts a small fraction of maxtime to use to cleanly shut down
> the PBS job. The default for this "reserve time" (see the Users Guide)
> is 10 seconds. So while before it was happily fitting 15-second jobs
> (the prior setting you had for maxwalltime) into (600-10) second
> slots, now (with maxwalltime increased to 10 minutes) it could not
> find any slots into which it could fit a 10 *minute* job.
> Unfortunately, at the moment, Swift just hangs, continuing to try to
> find a slot until the maxtime time runs out and the PBS jobs shut
> down. (There are "good" reasons for this "bad" behavior, which we need
> to fix)

A comment there:
Set the "maxtime" coaster parameter if you know that the queue you are
using has a limit to the time a job can have. Don't set it because your
jobs have a certain maxwalltime. It's purpose is to prevent the creation
of blocks that cannot be run on a given queue.




From marcin at galton.uchicago.edu  Tue Apr 27 12:00:39 2010
From: marcin at galton.uchicago.edu (Marcin Hitczenko)
Date: Tue, 27 Apr 2010 12:00:39 -0500 (CDT)
Subject: [Swift-user] Re: Set maxtime > maxwalltime or your script will hang
In-Reply-To: <17326438.669351272384369740.JavaMail.root@zimbra>
References: <17326438.669351272384369740.JavaMail.root@zimbra>
Message-ID: <44445.207.181.247.181.1272387639.squirrel@galton.uchicago.edu>

Hi,

> Marcin,
>
> ----- "Marcin Hitczenko" <marcin at galton.uchicago.edu> wrote:
>
>> Hi Mike,
>>
>> Two unrelated questions:
>>
>> 1. It seems then if I have 8 apptasks calls each of which is one hour
>> (maxwalltime is 01:00:00). I should be able to accomplish this by
>> submitting one PBS job for one node/ 8 cores and having each core run
>> a
>> different task. If I understand correctly, this should take on the
>> order
>> of 1 hour plus some extra. But, we would set maxwalltime to 60*60*8,
>> or 8
>> hours. Why should it run for 8 hours? I feel as I may still be
>> misunderstanding the distinctions.
>
> For this case, set maxwalltime to 01:00:00 (ie the estimated duration of
> each app() task) and maxtime to 3700 (1 hour with 100 secs "reserve")
>
> One caveat, though, for running on the Fusion cluster: I know that if you
> ask for the batch queue, it insists that you request at least 2 nodes; if
> you ask for one, your job is rejected. We need to test exactly how to
> configure coasters for 1 node on that system. You can try removing the
> queue element from your sites.xml, and see how it behaves. I need to do
> more testing on that system and post the results.

But even if we have 16 app() tasks and we call for 2 nodes, shouldn't
maxtime be more or less the same i.e. 3600+reserve? I guess my question
refers to maxtime=maxwalltime*N + reserve. Shouldn't N be the number of
waves or (# app() tasks)/(8cores/node*#nodes), rather than the # of app()
tasks?




>> 2. I submitted a job and would like to cancel it (because it has an
>> error). I use qdel to cancel the job, but within a few minutes a new
>> job
>> restarts. I take it this is swift retrying. How do I actually cancel
>> the
>> entire job? I have done qdel quite a few times, but jobs keep popping
>> up?

I am using the command: swift .... >& swift.out &, so I can't kill the
command directly, I don't think.

> I think the best way to clean up is to interrupt/kill the swift command
> with a ^C,
> and then it should clean up its PBS jobs. I think this has been working
> for me; let us know if that doesnt work for you.
>
> - Mike
>
>
>> Thanks,
>>
>> Marcin
>>
>>
>> > [cc'ing swift-user]
>> >
>> > Marcin,
>> >
>> > Quick answer: Since you changed the maxwalltime in tc.data to 10
>> minutes,
>> > change the "maxtime" setting in sites.xml to N times 10 minutes
>> *plus* 1
>> > minute. The "plus" is important.
>> >
>> > Long answer:
>> >
>> > Swift deducts a small fraction of maxtime to use to cleanly shut
>> down the
>> > PBS job. The default for this "reserve time" (see the Users Guide)
>> is 10
>> > seconds. So while before it was happily fitting 15-second jobs (the
>> prior
>> > setting you had for maxwalltime) into (600-10) second slots, now
>> (with
>> > maxwalltime increased to 10 minutes) it could not find any slots
>> into
>> > which it could fit a 10 *minute* job. Unfortunately, at the moment,
>> Swift
>> > just hangs, continuing to try to find a slot until the maxtime time
>> runs
>> > out and the PBS jobs shut down. (There are "good" reasons for this
>> "bad"
>> > behavior, which we need to fix)
>> >
>> > So bottom line: make maxtime some multiple of maxwalltime, and add a
>> bit
>> > to maxtime (say 1 minute, or at least 10 seconds). Note that since
>> > maxwalltime is just an *estimate* of how long you expect the apptask
>> to
>> > run for, this division is by necessity approximate. After an
>> apptask
>> > finishes on a coaster worker CPU, the CPU becomes free, and has
>> some
>> > varying amount of time left before the coaster worker expires. Then
>> the
>> > process repeats, and Swift again uses maxwalltime to see if there is
>> a
>> > coaster worker with at least maxwalltime remaining that can run the
>> next
>> > apptask.
>> >
>> > - Mike
>> >
>> > ps. How about if from now on, if you forget to cc swift-user on
>> these
>> > questions, then I will just cc the list on my replies. This has 2
>> > benefits: other users benefit from your questions and my answers,
>> and
>> > other swift developers and users can contribute more advice,
>> suggest
>> > betters approach, or correct me when I goof.
>> >
>> > You should join the swift-user list if you have not already done
>> so:
>> >   http://www.ci.uchicago.edu/swift/support/index.php
>> >
>> > Thanks!
>> >
>> > ----- "Marcin Hitczenko" <marcin at galton.uchicago.edu> wrote:
>> >
>> >> Hi Mike,
>> >>
>> >> Sorry to be such a pain, but I can't get the jobs to run again. I
>> am
>> >> still
>> >> running environment_setup.swift and have not made any changes to
>> the
>> >> pbscoast.xml file other than to change walltime. I am using the
>> same
>> >> tc.data file. I again run into the problem where it seems to run
>> for
>> >> a
>> >> minute, then fail and restart and it does so over and over. I
>> logged
>> >> into
>> >> the node and it didn't seem to be running my stuff. Also, the
>> output
>> >> files
>> >> are not being written.
>> >>
>> >> I am not sure why it would not work now, though I ran it and it
>> >> worked
>> >> before.
>> >>
>> >> Marcin
>> >>
>> >>
>> >>
>> >
>
> --
> Michael Wilde
> Computation Institute, University of Chicago
> Mathematics and Computer Science Division
> Argonne National Laboratory
>



From wilde at mcs.anl.gov  Tue Apr 27 12:05:40 2010
From: wilde at mcs.anl.gov (Michael Wilde)
Date: Tue, 27 Apr 2010 12:05:40 -0500
Subject: [Swift-user] Re: Set maxtime > maxwalltime or your script will hang
In-Reply-To: <44445.207.181.247.181.1272387639.squirrel@galton.uchicago.edu>
References: <17326438.669351272384369740.JavaMail.root@zimbra>
	<44445.207.181.247.181.1272387639.squirrel@galton.uchicago.edu>
Message-ID: <614A5D33-1D76-47F2-BAFE-1ACB627FCD03@mcs.anl.gov>


On Apr 27, 2010, at 12:00 PM, Marcin Hitczenko wrote:

> But even if we have 16 app() tasks and we call for 2 nodes, shouldn't
> maxtime be more or less the same i.e. 3600+reserve? I guess my question
> refers to maxtime=maxwalltime*N + reserve. Shouldn't N be the number of
> waves or (# app() tasks)/(8cores/node*#nodes), rather than the # of app()
> tasks?

Yes, sorry for the confusion - that is correct: by N I meant "waves", not jobs.

- Mike



From hategan at mcs.anl.gov  Tue Apr 27 12:12:27 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Tue, 27 Apr 2010 12:12:27 -0500
Subject: [Swift-user] Re: Set maxtime > maxwalltime or your script will
	hang
In-Reply-To: <44445.207.181.247.181.1272387639.squirrel@galton.uchicago.edu>
References: <17326438.669351272384369740.JavaMail.root@zimbra>
	<44445.207.181.247.181.1272387639.squirrel@galton.uchicago.edu>
Message-ID: <1272388347.6519.3.camel@localhost>

> > For this case, set maxwalltime to 01:00:00 (ie the estimated duration of
> > each app() task) and maxtime to 3700 (1 hour with 100 secs "reserve")
> >
> > One caveat, though, for running on the Fusion cluster: I know that if you
> > ask for the batch queue, it insists that you request at least 2 nodes; if
> > you ask for one, your job is rejected. We need to test exactly how to
> > configure coasters for 1 node on that system. You can try removing the
> > queue element from your sites.xml, and see how it behaves. I need to do
> > more testing on that system and post the results.
> 
> But even if we have 16 app() tasks and we call for 2 nodes, shouldn't
> maxtime be more or less the same i.e. 3600+reserve? I guess my question
> refers to maxtime=maxwalltime*N + reserve. Shouldn't N be the number of
> waves or (# app() tasks)/(8cores/node*#nodes), rather than the # of app()
> tasks?

No. Maxtime is something that is specific to the cluster not the jobs
you are running. It should be the maximum allowed time for a job in the
queue you are using. Actual coaster blocks may have a different
walltime, but they will never have a walltime larger than maxtime.



From wilde at mcs.anl.gov  Tue Apr 27 12:16:48 2010
From: wilde at mcs.anl.gov (Michael Wilde)
Date: Tue, 27 Apr 2010 12:16:48 -0500
Subject: [Swift-user] Set maxtime > maxwalltime or your script will hang
In-Reply-To: <1272386915.3670.2.camel@localhost>
References: <11830128.659071272366288457.JavaMail.root@zimbra>
	<1272386915.3670.2.camel@localhost>
Message-ID: <C1F02130-FA6C-4DCC-BDE6-54F16BAF0F28@mcs.anl.gov>


On Apr 27, 2010, at 11:48 AM, Mihael Hategan wrote:

> A comment there:
> Set the "maxtime" coaster parameter if you know that the queue you are
> using has a limit to the time a job can have. Don't set it because your
> jobs have a certain maxwalltime. It's purpose is to prevent the creation
> of blocks that cannot be run on a given queue.

Mihael, can you clarify this (and the related text from the User Guide)?

Users almost always want to get their jobs into a "good" queue, and hence
they usually feel compelled to set maxtime.  So more questions arise when
you try to figure out how to do this:

- How does coasters fill blocks?

- What LRM wall time will be used when maxtime is not specified? 
  (i.e, how many jobs will coasters place in a block)?

- Is coasters committing more jobs to specific blocks (and hence specific sites)
  than the block has cores,  when it computes its schedules?
  Or just estimating what it may need? 





From hategan at mcs.anl.gov  Tue Apr 27 12:30:40 2010
From: hategan at mcs.anl.gov (Mihael Hategan)
Date: Tue, 27 Apr 2010 12:30:40 -0500
Subject: [Swift-user] Set maxtime > maxwalltime or your script will hang
In-Reply-To: <C1F02130-FA6C-4DCC-BDE6-54F16BAF0F28@mcs.anl.gov>
References: <11830128.659071272366288457.JavaMail.root@zimbra>
	<1272386915.3670.2.camel@localhost>
	<C1F02130-FA6C-4DCC-BDE6-54F16BAF0F28@mcs.anl.gov>
Message-ID: <1272389440.6945.8.camel@localhost>

On Tue, 2010-04-27 at 12:16 -0500, Michael Wilde wrote:
> On Apr 27, 2010, at 11:48 AM, Mihael Hategan wrote:
> 
> > A comment there:
> > Set the "maxtime" coaster parameter if you know that the queue you are
> > using has a limit to the time a job can have. Don't set it because your
> > jobs have a certain maxwalltime. It's purpose is to prevent the creation
> > of blocks that cannot be run on a given queue.
> 
> Mihael, can you clarify this (and the related text from the User Guide)?
> 
> Users almost always want to get their jobs into a "good" queue, and hence
> they usually feel compelled to set maxtime.  So more questions arise when
> you try to figure out how to do this:
> 
> - How does coasters fill blocks?

Largest job in the smallest space that can fit it. Then repeat.

> 
> - What LRM wall time will be used when maxtime is not specified? 
>   (i.e, how many jobs will coasters place in a block)?

That is controlled by the overallocation numbers. Here's the role of
maxtime:

blockTime = calculateFromJobsAndOverallocationSettings();
if (blockTime > maxtime) {
  blockTime = maxtime;
}

So you shouldn't try to use "maxtime" to control the block walltime.
That is not its purpose. And I would rename it to have a clearer name,
except I can't think of a better name than "maxtime" for what it does.
Or maybe "neverExceedTime".

> 
> - Is coasters committing more jobs to specific blocks (and hence specific sites)
>   than the block has cores,  when it computes its schedules?

It depends on whether you want parallelism or efficiency. The default
values tries for parallelism, so it will try to allocate as many cores
as you have jobs. So typically you won't see more cores allocated than
jobs, unless you are hitting some granularity.

>   Or just estimating what it may need?

It is estimating what it may need.