2008-08-06 18:36:30 +00:00
|
|
|
// This file is part of BOINC.
|
2005-08-04 03:50:04 +00:00
|
|
|
// http://boinc.berkeley.edu
|
2008-08-06 18:36:30 +00:00
|
|
|
// Copyright (C) 2008 University of California
|
2005-08-04 03:50:04 +00:00
|
|
|
//
|
2008-08-06 18:36:30 +00:00
|
|
|
// BOINC is free software; you can redistribute it and/or modify it
|
|
|
|
// under the terms of the GNU Lesser General Public License
|
|
|
|
// as published by the Free Software Foundation,
|
|
|
|
// either version 3 of the License, or (at your option) any later version.
|
2005-08-04 03:50:04 +00:00
|
|
|
//
|
2008-08-06 18:36:30 +00:00
|
|
|
// BOINC is distributed in the hope that it will be useful,
|
2005-08-04 03:50:04 +00:00
|
|
|
// but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
// MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
|
|
|
|
// See the GNU Lesser General Public License for more details.
|
|
|
|
//
|
2008-08-06 18:36:30 +00:00
|
|
|
// You should have received a copy of the GNU Lesser General Public License
|
|
|
|
// along with BOINC. If not, see <http://www.gnu.org/licenses/>.
|
2005-08-04 03:50:04 +00:00
|
|
|
|
- scheduler: add <workload_sim> config option.
If set, the scheduler will use EDF simulation,
together with the in-progress workload reported by the client,
to avoid sending results that
1) will miss their deadline, or
2) will cause an in-progress result to miss its deadline, or
3) will make an in-progress result miss its deadline
by more than is already predicted.
If this option is not set, or if the client request doesn't
include a workload description (i.e. the client is old)
use the existing approach, which assumes there's no workload.
NOTE: this is experimental. Production projects should not use it.
- EDF sim: write debug stuff to stderr instead of stdout
- Account manager:
- if an account is detach_when_done, set dont_request_more_work
- check done_request_more_work even for first-time projects
- update_uotd: generate a file for use by Google gadget
- user_links(): use full URLs (so can use in Google gadget)
client/
acct_mgr.C
work_fetch.C
html/
inc/
uotd.inc
util.inc
user/
uotd_gadget.php (new)
sched/
Makefile.am
edf_sim.C
sched_config.C,h
sched_resend.C
sched_send.C,h
server_types.C,h
svn path=/trunk/boinc/; revision=12639
2007-05-10 21:50:52 +00:00
|
|
|
// scheduler code related to sending "lost" work
|
|
|
|
// (i.e. results we sent to the host, but which they're not reporting)
|
|
|
|
//
|
|
|
|
// TODO:
|
|
|
|
// - make sure result is still needed (no canonical result yet)
|
|
|
|
// - don't send if project has been reset since first send;
|
|
|
|
// this result may have been the cause of reset
|
|
|
|
// (need to pass last reset time from client)
|
2005-08-04 03:50:04 +00:00
|
|
|
|
2009-03-19 16:35:35 +00:00
|
|
|
#include "config.h"
|
|
|
|
|
2008-02-27 23:26:38 +00:00
|
|
|
#include <cstdlib>
|
|
|
|
#include <cstring>
|
|
|
|
#include <string>
|
|
|
|
|
2005-08-04 03:50:04 +00:00
|
|
|
#include "error_numbers.h"
|
|
|
|
|
2009-08-10 04:49:02 +00:00
|
|
|
#include "sched_main.h"
|
2005-08-04 03:50:04 +00:00
|
|
|
#include "sched_config.h"
|
2009-03-19 16:35:35 +00:00
|
|
|
#include "sched_locality.h"
|
2005-08-04 03:50:04 +00:00
|
|
|
#include "sched_msgs.h"
|
|
|
|
#include "sched_send.h"
|
2009-03-19 16:35:35 +00:00
|
|
|
#include "sched_shmem.h"
|
|
|
|
#include "sched_util.h"
|
|
|
|
#include "sched_version.h"
|
2009-08-10 04:49:02 +00:00
|
|
|
#include "sched_types.h"
|
2005-08-04 03:58:00 +00:00
|
|
|
|
|
|
|
#include "sched_resend.h"
|
2005-08-04 03:50:04 +00:00
|
|
|
|
|
|
|
|
|
|
|
#ifdef _USING_FCGI_
|
2008-09-09 19:10:42 +00:00
|
|
|
#include "boinc_fcgi.h"
|
2005-08-04 03:50:04 +00:00
|
|
|
#endif
|
|
|
|
|
- scheduler: add <workload_sim> config option.
If set, the scheduler will use EDF simulation,
together with the in-progress workload reported by the client,
to avoid sending results that
1) will miss their deadline, or
2) will cause an in-progress result to miss its deadline, or
3) will make an in-progress result miss its deadline
by more than is already predicted.
If this option is not set, or if the client request doesn't
include a workload description (i.e. the client is old)
use the existing approach, which assumes there's no workload.
NOTE: this is experimental. Production projects should not use it.
- EDF sim: write debug stuff to stderr instead of stdout
- Account manager:
- if an account is detach_when_done, set dont_request_more_work
- check done_request_more_work even for first-time projects
- update_uotd: generate a file for use by Google gadget
- user_links(): use full URLs (so can use in Google gadget)
client/
acct_mgr.C
work_fetch.C
html/
inc/
uotd.inc
util.inc
user/
uotd_gadget.php (new)
sched/
Makefile.am
edf_sim.C
sched_config.C,h
sched_resend.C
sched_send.C,h
server_types.C,h
svn path=/trunk/boinc/; revision=12639
2007-05-10 21:50:52 +00:00
|
|
|
// Assign a new deadline for the result;
|
|
|
|
// if it's not likely to complete by this time, return nonzero.
|
|
|
|
// TODO: EXPLAIN THE FORMULA FOR NEW DEADLINE
|
2005-08-04 03:50:04 +00:00
|
|
|
//
|
|
|
|
static int possibly_give_result_new_deadline(
|
2008-12-15 21:14:32 +00:00
|
|
|
DB_RESULT& result, WORKUNIT& wu, BEST_APP_VERSION& bav
|
2005-08-04 03:50:04 +00:00
|
|
|
) {
|
|
|
|
const double resend_frac = 0.5; // range [0, 1)
|
2007-11-30 23:02:55 +00:00
|
|
|
int now = time(0);
|
|
|
|
int result_report_deadline = now + (int)(resend_frac*(result.report_deadline - result.sent_time));
|
2005-08-04 03:50:04 +00:00
|
|
|
|
|
|
|
if (result_report_deadline < result.report_deadline) {
|
|
|
|
result_report_deadline = result.report_deadline;
|
|
|
|
}
|
2007-11-30 23:02:55 +00:00
|
|
|
if (result_report_deadline > now + wu.delay_bound) {
|
|
|
|
result_report_deadline = now + wu.delay_bound;
|
2005-08-04 03:50:04 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
// If infeasible, return without modifying result
|
|
|
|
//
|
2008-12-15 21:14:32 +00:00
|
|
|
if (estimate_duration(wu, bav) > result_report_deadline-now) {
|
2008-08-19 03:00:17 +00:00
|
|
|
if (config.debug_resend) {
|
2009-01-15 20:23:20 +00:00
|
|
|
log_messages.printf(MSG_NORMAL,
|
|
|
|
"[resend] [RESULT#%d] [HOST#%d] not resending lost result: can't complete in time\n",
|
2008-12-15 21:14:32 +00:00
|
|
|
result.id, g_reply->host.id
|
2008-08-19 03:00:17 +00:00
|
|
|
);
|
|
|
|
}
|
2005-08-04 03:50:04 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
// update result with new report time and sent time
|
|
|
|
//
|
2008-08-19 03:00:17 +00:00
|
|
|
if (config.debug_resend) {
|
2009-01-15 20:23:20 +00:00
|
|
|
log_messages.printf(MSG_NORMAL,
|
|
|
|
"[resend] [RESULT#%d] [HOST#%d] %s report_deadline (resend lost work)\n",
|
2008-12-15 21:14:32 +00:00
|
|
|
result.id, g_reply->host.id,
|
2008-08-19 03:00:17 +00:00
|
|
|
result_report_deadline==result.report_deadline?"NO update to":"Updated"
|
|
|
|
);
|
|
|
|
}
|
2007-11-30 23:02:55 +00:00
|
|
|
result.sent_time = now;
|
2005-08-04 03:50:04 +00:00
|
|
|
result.report_deadline = result_report_deadline;
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2007-11-30 23:02:55 +00:00
|
|
|
// resend any jobs that:
|
|
|
|
// 1) we already sent to this host;
|
|
|
|
// 2) are still in progress (i.e. haven't timed out) and
|
|
|
|
// 3) aren't present on the host
|
|
|
|
// Return true if there were any such jobs
|
|
|
|
//
|
2008-12-19 18:14:02 +00:00
|
|
|
bool resend_lost_work() {
|
2011-06-06 03:40:42 +00:00
|
|
|
SCHED_DB_RESULT result;
|
2005-08-04 03:50:04 +00:00
|
|
|
std::vector<DB_RESULT>results;
|
|
|
|
unsigned int i;
|
|
|
|
char buf[256];
|
2007-11-30 23:02:55 +00:00
|
|
|
char warning_msg[256];
|
2005-08-04 03:50:04 +00:00
|
|
|
bool did_any = false;
|
2007-11-30 23:02:55 +00:00
|
|
|
int num_eligible_to_resend=0;
|
2005-08-04 03:50:04 +00:00
|
|
|
int num_resent=0;
|
2008-03-27 18:25:29 +00:00
|
|
|
BEST_APP_VERSION* bavp;
|
2005-08-04 03:50:04 +00:00
|
|
|
int retval;
|
|
|
|
|
|
|
|
sprintf(buf, " where hostid=%d and server_state=%d ",
|
2008-12-19 18:14:02 +00:00
|
|
|
g_reply->host.id, RESULT_SERVER_STATE_IN_PROGRESS
|
2005-08-04 03:50:04 +00:00
|
|
|
);
|
|
|
|
while (!result.enumerate(buf)) {
|
2010-05-11 19:50:14 +00:00
|
|
|
if (!work_needed(false)) {
|
|
|
|
result.end_enumerate();
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2005-08-04 03:50:04 +00:00
|
|
|
bool found = false;
|
2008-12-19 18:14:02 +00:00
|
|
|
for (i=0; i<g_request->other_results.size(); i++) {
|
|
|
|
OTHER_RESULT& orp = g_request->other_results[i];
|
2009-06-01 22:15:14 +00:00
|
|
|
if (!strcmp(orp.name, result.name)) {
|
2005-08-04 03:50:04 +00:00
|
|
|
found = true;
|
2007-11-30 23:02:55 +00:00
|
|
|
break;
|
2005-08-04 03:50:04 +00:00
|
|
|
}
|
|
|
|
}
|
2007-11-30 23:02:55 +00:00
|
|
|
if (found) continue;
|
|
|
|
|
|
|
|
num_eligible_to_resend++;
|
2008-08-19 03:00:17 +00:00
|
|
|
if (config.debug_resend) {
|
2009-01-15 20:23:20 +00:00
|
|
|
log_messages.printf(MSG_NORMAL,
|
|
|
|
"[resend] [HOST#%d] found lost [RESULT#%d]: %s\n",
|
2008-12-19 18:14:02 +00:00
|
|
|
g_reply->host.id, result.id, result.name
|
2008-08-19 03:00:17 +00:00
|
|
|
);
|
|
|
|
}
|
2007-11-30 23:02:55 +00:00
|
|
|
|
|
|
|
DB_WORKUNIT wu;
|
|
|
|
retval = wu.lookup_id(result.workunitid);
|
|
|
|
if (retval) {
|
2008-02-21 21:00:58 +00:00
|
|
|
log_messages.printf(MSG_CRITICAL,
|
2007-11-30 23:02:55 +00:00
|
|
|
"[HOST#%d] WU not found for [RESULT#%d]\n",
|
2008-12-19 18:14:02 +00:00
|
|
|
g_reply->host.id, result.id
|
2005-08-04 03:50:04 +00:00
|
|
|
);
|
2007-11-30 23:02:55 +00:00
|
|
|
continue;
|
|
|
|
}
|
2005-08-04 03:50:04 +00:00
|
|
|
|
2011-02-24 19:30:43 +00:00
|
|
|
APP* app = ssp->lookup_app(wu.appid);
|
- server: change the following from per-host to per-(host, app version):
- daily quota mechanism
- reliable mechanism (accelerated retries)
- "trusted" mechanism (adaptive replication)
- scheduler: enforce host scale probation only for apps with
host_scale_check set.
- validator: do scale probation on invalid results
(need this in addition to error and timeout cases)
- feeder: update app version scales every 10 min, not 10 sec
- back-end apps: support --foo as well as -foo for options
Notes:
- If you have, say, cuda, cuda23 and cuda_fermi plan classes,
a host will have separate quotas for each one.
That means it could error out on 100 jobs for cuda_fermi,
and when its quota goes to zero,
error out on 100 jobs for cuda23, etc.
This is intentional; there may be cases where one version
works but not the others.
- host.error_rate and host.max_results_day are deprecated
TODO:
- the values in the app table for limits on jobs in progress etc.
should override rather than config.xml.
Implementation notes:
scheduler:
process_request():
read all host_app_versions for host at start;
Compute "reliable" and "trusted" for each one.
write modified records at end
get_app_version():
add "reliable_only" arg; if set, use only reliable versions
skip over-quota versions
Multi-pass scheduling: if have at least one reliable version,
do a pass for jobs that need reliable,
and use only reliable versions.
Then clear best_app_versions cache.
Score-based scheduling: for need-reliable jobs,
it will pick the fastest version,
then give a score bonus if that version happens to be reliable.
When get back a successful result from client:
increase daily quota
When get back an error result from client:
impose scale probation
decrease daily quota if not aborted
Validator:
when handling a WU, create a vector of HOST_APP_VERSION
parallel to vector of RESULT.
Pass it to assign_credit_set().
Make copies of originals so we can update only modified ones
update HOST_APP_VERSION error rates
Transitioner:
decrease quota on timeout
svn path=/trunk/boinc/; revision=21181
2010-04-15 03:13:56 +00:00
|
|
|
bavp = get_app_version(wu, false, false);
|
2008-03-27 18:25:29 +00:00
|
|
|
if (!bavp) {
|
2008-02-21 21:00:58 +00:00
|
|
|
log_messages.printf(MSG_CRITICAL,
|
2009-03-05 22:12:21 +00:00
|
|
|
"[HOST#%d] can't resend [RESULT#%d]: no app version for %s\n",
|
|
|
|
g_reply->host.id, result.id, app->name
|
2007-11-30 23:02:55 +00:00
|
|
|
);
|
|
|
|
continue;
|
|
|
|
}
|
2005-08-04 03:50:04 +00:00
|
|
|
|
2007-11-30 23:02:55 +00:00
|
|
|
// If time is too close to the deadline,
|
|
|
|
// or we already have a canonical result,
|
|
|
|
// or WU error flag is set,
|
|
|
|
// then don't bother to resend this result.
|
|
|
|
// Instead make it time out right away
|
|
|
|
// so that the transitioner does 'the right thing'.
|
|
|
|
//
|
|
|
|
if (
|
2011-02-24 19:30:43 +00:00
|
|
|
wu.error_mask
|
|
|
|
|| wu.canonical_resultid
|
|
|
|
|| wu_is_infeasible_fast(
|
|
|
|
wu, result.server_state, result.priority, result.report_deadline,
|
|
|
|
*app, *bavp
|
|
|
|
)
|
|
|
|
|| possibly_give_result_new_deadline(result, wu, *bavp)
|
2007-11-30 23:02:55 +00:00
|
|
|
) {
|
2008-08-19 03:00:17 +00:00
|
|
|
if (config.debug_resend) {
|
2009-01-15 20:23:20 +00:00
|
|
|
log_messages.printf(MSG_NORMAL,
|
|
|
|
"[resend] [HOST#%d][RESULT#%d] not needed or too close to deadline, expiring\n",
|
2008-12-19 18:14:02 +00:00
|
|
|
g_reply->host.id, result.id
|
2008-08-19 03:00:17 +00:00
|
|
|
);
|
|
|
|
}
|
2007-11-30 23:02:55 +00:00
|
|
|
result.report_deadline = time(0)-1;
|
2011-02-15 22:07:14 +00:00
|
|
|
retval = result.mark_as_sent(result.server_state, config.report_grace_period);
|
2005-08-04 03:50:04 +00:00
|
|
|
if (retval) {
|
2008-02-21 21:00:58 +00:00
|
|
|
log_messages.printf(MSG_CRITICAL,
|
2010-11-08 17:51:57 +00:00
|
|
|
"resend_lost_work: can't update result deadline: %s\n",
|
|
|
|
boincerror(retval)
|
2005-08-04 03:50:04 +00:00
|
|
|
);
|
|
|
|
continue;
|
|
|
|
}
|
|
|
|
|
2011-06-06 03:40:42 +00:00
|
|
|
retval = update_wu_on_send(
|
|
|
|
wu, result.report_deadline + config.report_grace_period,
|
|
|
|
*app, *bavp
|
2011-02-15 22:07:14 +00:00
|
|
|
);
|
2007-11-30 23:02:55 +00:00
|
|
|
if (retval) {
|
2008-02-21 21:00:58 +00:00
|
|
|
log_messages.printf(MSG_CRITICAL,
|
2010-11-08 17:51:57 +00:00
|
|
|
"resend_lost_result: can't update WU transition time: %s\n",
|
|
|
|
boincerror(retval)
|
2005-08-04 03:50:04 +00:00
|
|
|
);
|
|
|
|
continue;
|
|
|
|
}
|
2007-11-30 23:02:55 +00:00
|
|
|
sprintf(warning_msg,
|
2010-03-03 19:29:23 +00:00
|
|
|
"Didn't resend lost task %s (expired)", result.name
|
2007-11-30 23:02:55 +00:00
|
|
|
);
|
2010-06-15 17:56:30 +00:00
|
|
|
g_reply->insert_message(warning_msg, "low");
|
2007-11-30 23:02:55 +00:00
|
|
|
} else {
|
2009-03-03 16:38:54 +00:00
|
|
|
retval = add_result_to_reply(result, wu, bavp, false);
|
2005-08-04 03:50:04 +00:00
|
|
|
if (retval) {
|
2008-02-21 21:00:58 +00:00
|
|
|
log_messages.printf(MSG_CRITICAL,
|
2005-08-04 03:50:04 +00:00
|
|
|
"[HOST#%d] failed to send [RESULT#%d]\n",
|
2008-12-19 18:14:02 +00:00
|
|
|
g_reply->host.id, result.id
|
2005-08-04 03:50:04 +00:00
|
|
|
);
|
|
|
|
continue;
|
|
|
|
}
|
2010-03-03 19:29:23 +00:00
|
|
|
sprintf(warning_msg, "Resent lost task %s", result.name);
|
2010-06-15 17:56:30 +00:00
|
|
|
g_reply->insert_message(warning_msg, "low");
|
2005-08-04 03:50:04 +00:00
|
|
|
num_resent++;
|
|
|
|
did_any = true;
|
2009-05-15 21:12:46 +00:00
|
|
|
|
2009-06-01 22:15:14 +00:00
|
|
|
if (g_wreq->njobs_sent >= config.max_wus_to_send) {
|
2009-05-15 21:12:46 +00:00
|
|
|
result.end_enumerate();
|
|
|
|
break;
|
|
|
|
}
|
2005-08-04 03:50:04 +00:00
|
|
|
}
|
|
|
|
}
|
2007-05-14 15:21:38 +00:00
|
|
|
|
2008-08-19 03:00:17 +00:00
|
|
|
if (num_eligible_to_resend && config.debug_resend) {
|
2009-01-15 20:23:20 +00:00
|
|
|
log_messages.printf(MSG_NORMAL,
|
2009-05-15 21:12:46 +00:00
|
|
|
"[resend] [HOST#%d] %d lost results, resent %d\n",
|
|
|
|
g_reply->host.id, num_eligible_to_resend, num_resent
|
2005-08-04 03:50:04 +00:00
|
|
|
);
|
|
|
|
}
|
|
|
|
|
|
|
|
return did_any;
|
|
|
|
}
|
|
|
|
|
2005-09-13 09:01:56 +00:00
|
|
|
const char *BOINC_RCSID_3be23838b4="$Id$";
|