Hi Raul, Quoting [Raul Sirvent] (Nov 07 2005):

Hi. I would like to give our opinion (from GRID superscalar's point of view, as a use case of SAGA) about this monitoring API.

The way it should be used is almost clear to me (it is very similar to the one defined in GAT). What I find missing is a way of waiting for callbacks to arrive, without having to wait for the task to finish.

you can do a while (1) { // callbacks will eventually arrive } or in fact anything else, as the callbacks are to be deliverd asynchroneously. No need to care about if the task is really finished

Of course I'm thinking in a Globus fashion (calls like globus_poll and globus_poll_blocking).

Ah, I am having a Deja Vu here :-) I found in one of our earlier discussions: Quoting [Raul Sirvent] :

We want to wake up every time a state change comes. In your case you are implementing a polling model but we are interested in having a event-driven model. Something like:

submit_first_jobs ();

while ( ! all_jobs_done ) { wait_for_notification ();

// will submit new jobs if // dependencies are solved treat_notification (); }

I don't see how to do that with the SAGA Task interface.

I think, in SAGA that would look like: // run jobs, and add notification callbacks to all of // them submit_first_jobs (); while ( ! all_jobs_done ) { // idle somewhat. Maybe a callback gets called :-) nanosleep (1); // A callback gets called in a different therad, so we // don't see it here, unless it sets some flags. // Well, assume that this it was the CB does, and that // treat_notification () is checking these flags, and // acts accordingly, e.g. submits new jobs treat_notification (); } So the code looks actually exactly like yours, w/o an explicit wait/poll. the CB is asynchroneous and in a different thread, and you can set a flag there, and poll for that, to notify your main thread. Now, that description obviously hold only for implementations which implement CB really asynchroneously, meaning in threads. If that is not the case, we would indeed need a explicit polling point, such as in GAT.

I remember something similar in GAT named.. ServiceActions?

exactly. That is nothing else than the wait_for_notification, or the globus poll.

Its purpose was to give the CPU to the GAT environment in order to do this kind of things (executing the functions to handle de callbacks, for instance). Is there something similar in SAGA?

Its not needed, as the second thread used for the notification can get CPU as needed. Hmm, I hope that makes sense :-) Cheers, Andre.

Cheers. Raul.

Rosa M. Badia wrote:

...

Raul,

em sembla que aixo te a veure amb els callbacks. Li dones una ullada?

Rosa

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

Subject: [saga-rg] monitoring, again... From: Andre Merzky Date: Sun, 6 Nov 2005 19:48:50 -0600 To: Simple API for Grid Applications WG

To: Simple API for Grid Applications WG

Hi,

as has been discussed recently, we want to have a simple way to add callbacks to SAGA. The main use case for this would be for notification on status changes for tasks and jobs.

Attached is the re-surrected monitoring package as a proposal to achieve this. Please note that the package mentiones steering at various places (mostly commented out) - that is as of yet not part of the proposal, but should seen as a possible path for future extension to steering.

For simplicity, a code example is included. More feedback to the topic is welcome. I Cc to the GridRPC folx, as they are probably interested in that.

Note that the example monitors not the task status, but some other metric provided by that implementation (bytes written).

Cheers, Andre.

+-------------------------------------------------------------+ Examples:

monitoring example: monitor a write task ---------------------------------------- short version: --------------

class write_metric_cb : public saga::metric::callback { public: void callback (saga::metric & m) { std::cout << atoi ( m.get_attribute ("value") ) << " "; } };

int main (int argc, char** argv) { ssize_t len = 0; std::string str ("Hello SAGA\n"); std::string url (argv[1]);

saga::file f (url); saga::task t = f.task.write (str, &len); saga::metric m = t.get_metric ("Written");

write_metric_cb cb; m.add_callback (cb);

t.wait (); }

monitoring example: monitor a write task ---------------------------------------- long annotated version: -----------------------

// this example shows how monitoring a task can be // implemented class write_metric_cb : public saga::metric::callback { private: saga::task t;

public: write_metric_cb (const saga::task & _t) { t = _t; }

void callback (saga::metric & m) { int len = atoi ( m.get_attribute ("value") ); std::cout << "bytes written: " << len << std::endl; std::cout << "task status: " << t.get_status () << std::endl; } };

int main (int argc, char** argv) { ssize_t len = 0; std::string str ("Hello SAGA\n"); std::string url (argv[1]);

saga::file f (url); saga::task t = f.task.write (str, &len); saga::metric m = t.get_metric ("Written");

// assume that this m is a discreet metric indicating // the number of bytes already written. In general, // the list of metric names has to be searched for an // interesting metric if name is not known.

// add the callback write_metric_cb cb; m.add_callback (cb);

// wait until task is done, and give cb chance to get // called a couple of times t.wait (); } +-------------------------------------------------------------+

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

+-------------------------------------------------------------+

# # ## ## ### # # # ##### ### #### # # # ### # # # # # # ## # # # # # # # # ## # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # # #### # # # # # ### # # # # # ## # # # # # # # # ## # # # # ### # # # # ### # # # # # ###

## # # ## ### # # # # # ### # ##### # # ##### ###### ###### ##### # # # #### # # # # # # # ## # # # ##### # ##### ##### # # # # # # # # # # # ##### # # # # # ### # # # # # # # # # ## # # ##### # ###### ###### # # # # # ####

+-------------------------------------------------------------+

Summary: ========

The ability to query Grid entities about state is requested in several SAGA use cases. Also, the SAGA Task model incorporates a certain amount of task monitoring.

This package definition approaches the problem space of monitoring to unify the various usage patterns (see details), and to transparently incorporate SAGA task monitoring.

A closely related topic is Steering, which is not really seen independently from Monitoring: in the SAGA approach, a future Steering may extend Monitoring by the ability to push values back to the Monitoring source.

+-------------------------------------------------------------+

Specification: ==============

package SAGA version 0.1 {

package Monitoring {

class Metric implements-all Attribute {

// need to be derived from for callbacks class CallBack { void callback (in Metric metric); }

// add a callback, which gets active whenever // the metric changes addCallBack (in CallBack cb, out int cookie); removeCallBack (in int cb);

// steering: update the metric value // update (in string value); }

interface Monitorable { // introspection getNames (out array names);

// get hook for monitoring/steering getMetric (in String name, out Metric metric);

// set hook for steering addMetric (out Metric metric); } } }

+-------------------------------------------------------------+

#ifndef SHORT

Details: ========

class Metric

The fundamenta object introduced in this package is a Metric. A metric representas an descreet or continuous observable, which can be read, write, or read/writable. A readable observable corresponds to classical monitoring, a writable observable corresponds to steering.

The approach is severely limited by the use of saga attributes for the description of a Metric, as these are only defined in terms of string keys and values. An extension of the attribute definition by typed and complex values will greatly improve the usability of this package, but will also challenge its semantic simplicity.

The metric MUST provide access to following attributes:

"Name" : short human readable name "file.transfer.progress "Description": extensive human readable description "This metric gives the status of an ongoing file transfer as percent completed." "Freq": Discreet or Continuous "CONTINUOUS" "Mode": "Read", "Write", "ReadWrite" or "Final" "Read" "Unit": Unit of values "percent (%)" "Type": "String", "Int", "Float" etc. "Float" "Value": value of the metric "20.5"

The name of the metric must be unique, as the get_metric call needs to be able to identify the metric to return.

A writable metric can be updated, and hence provides remote steering capabilities. However, that is currently not part of the spec, but noted here as a path for future extension of the intertface.

+-------------------------------------------------------------+

Examples: =========

monitoring example: monitor a write task ----------------------------------------

// this example shows how monitoring a task can be // implemented class write_metric_cb : public saga::metric::callback { private: saga::task t;

public: write_metric_cb (const saga::task & _t) { t = _t; }

void callback (saga::metric & m) { int len = atoi ( m.get_attribute ("value") ); std::cout << "bytes written: " << len << std::endl; std::cout << "task status: " << t.get_status () << std::endl; } };

int main (int argc, char** argv) { ssize_t len = 0; std::string str ("Hello SAGA\n"); std::string url (argv[1]);

saga::file f (url); saga::task t = f.task.write (str, &len); saga::metric m = t.get_metric ("Progress");

// assume that this m is a discreet metric indicating // the number of bytes already written. In general, // the list of metric names has to be searched for an // interesting metric.

// add the callback write_metric_callback cb (t); m.add_callback (cb);

// wait until task is done, and give cb chance to get // called a couple of times t.wait (); }

steering example: steer a remote job ------------------------------------

// this example is concerned with steering (metric is writable). // However, Steering is currentlu NOT part of the SAGA spec. class application_observer_cb : public saga::metric::callback { private: saga::task t;

public: void callback (saga::metric & m) { int val = atoi ( m.get_attribute ("value") ); std::cout << val << " is the new value." << std::endl; } };

int main (int argc, char** argv) { saga::job_service js;

saga::job j = js.run ("remote.host.net", "my_remote_application");

saga::metric m = job.get_metric ("param_1");

// assume that m is a discreet metric representing // a integer parameter for the remote application. application_observer_cb cb; m.add_callback (cb);

for ( int i = 0; i < 10; i++ ) { m.update (std::string (i)); // callback should get called NOW + latency // if param_1 is read only, update would return an error. sleep (1); } }

steering example: BE a steerable job ------------------------------------

// this example assumes that a job representing THIS application // (self) allows to add a metric. However, that is currently // NOT part of the SAGA spec. class application_observable_cb : public saga::metric::callback { private: int & i;

public: application_observable_cb (int & _i) { i = _i; } void callback (saga::metric & m) { int val = atoi ( m.get_attribute ("value") ); i = val; std::cout << "new value: " << i << std::endl; } };

int main (int argc, char** argv) { saga::job j = theSession.get_self ();

saga::metric m ("TestMetric", // name "This is a test metric", // description "Discreet", // frequency "ReadWrite", // mode "", // no unit "Int", // type "0"); // initial value

j.add_metric (m);

// a remote jobs can steer us for 100 seconds sleep (100); }

+---------------------------------------------------------------+

Notes: ======

- possible deviation: allow only one CB per metric: no add/remove, but set/reset CB

+-------------------------------------------------------------+

#endif // SHORT

-- +-----------------------------------------------------------------+ | Andre Merzky | phon: +31 - 20 - 598 - 7759 | | Vrije Universiteit Amsterdam (VU) | fax : +31 - 20 - 598 - 7653 | | Dept. of Computer Science | mail: merzky@cs.vu.nl | | De Boelelaan 1083a | www: http://www.merzky.net | | 1081 HV Amsterdam, Netherlands | | +-----------------------------------------------------------------+