While checking the logic today I noticed that the scaling of
config.max_wus_to_send first added in changeset 18255 wasn't
applied in sched_resend.cpp so the resends are likely to be
very limited. It seems to me that the scaling ought to be used,
and is practical because handle_request does send_work_setup()
before calling resend_lost_work(). Suggested change:
Index: sched_resend.cpp
===================================================================
--- sched_resend.cpp (revision 22314)
+++ sched_resend.cpp (working copy)
@@ -210,7 +210,7 @@
num_resent++;
did_any = true;
- if (g_wreq->njobs_sent >= config.max_wus_to_send) {
+ if (g_wreq->njobs_sent >= g_wreq->max_jobs_per_rpc) {
result.end_enumerate();
break;
}
--
Joe
On 11 Aug 2010 at 15:04, David wrote:
> I checked in the server change
> (if a request reports already-reported results,
> always resend lost results).
>
> This will make its way into s...@home in a week or two.
>
> -- David
_______________________________________________
boinc_dev mailing list
[email protected]
http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev
To unsubscribe, visit the above URL and
(near bottom of page) enter your email address.
