Not started by deadline - canceled ???

Message boards : Number crunching : Not started by deadline - canceled ???

Author	Message
alan Send message Joined: 18 Feb 00 Posts: 125 Credit: 311,600 RAC: 390	Message 1260705 - Posted: 15 Jul 2012, 17:17:10 UTC
	What does this message mean, as found in this work unit? I'm sure it didn't say this to start with, but can't remember what was there. It may have been "aborted", but I'm not sure. ____________
	ID: 1260705 ·

Fred J. Verster Volunteer tester Send message Joined: 21 Apr 04 Posts: 3232 Credit: 31,585,541 RAC: 1	Message 1260871 - Posted: 15 Jul 2012, 22:03:02 UTC - in response to Message 1260705. Last modified: 15 Jul 2012, 22:04:24 UTC
	What does this message mean, as found in this work unit? I'm sure it didn't say this to start with, but can't remember what was there. It may have been "aborted", but I'm not sure. Deadline is unclear. ____________ Knight Who Says Ni N!, OUT numbered.................
	ID: 1260871 ·

Horacio Send message Joined: 14 Jan 00 Posts: 523 Credit: 30,387,944 RAC: 130,288	Message 1260880 - Posted: 15 Jul 2012, 22:13:25 UTC
	The deadline of that task is on Aug. 24 but it was cancelled on July 6, with "Exit status : 200 (0xc8) EXIT_UNSTARTED_LATE" and "exit status: Aborted by the user"... I think that for some reason that host was set to some date in the future and then the BOINC client canceled the WU thinking that was too late to start it... ____________
	ID: 1260880 ·

alan Send message Joined: 18 Feb 00 Posts: 125 Credit: 311,600 RAC: 390	Message 1260888 - Posted: 15 Jul 2012, 22:54:11 UTC
	Horacio, your theory is plausible. But I'd expect an error like "timed out" if it had detected that it had run out of time. The guilty computer is a Linux box, so perhaps the Linux clients report different errors, and change them after a while. ____________
	ID: 1260888 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7459 Credit: 39,674,175 RAC: 33,868	Message 1260900 - Posted: 15 Jul 2012, 23:25:47 UTC - in response to Message 1260888.
	Horacio, your theory is plausible. But I'd expect an error like "timed out" if it had detected that it had run out of time. The guilty computer is a Linux box, so perhaps the Linux clients report different errors, and change them after a while. The clients all report the same codes (though there's some extra detail from v7.0.28 onwards). The behaviour - cancelling a task if it hasn't been started by the time the deadline is reached - has been around for a long time (since sometime in the v6.6.xx range, IIRC): but the specific message "Not started by deadline - canceled" is new since last week. Before that, they would have been displayed as "Aborted by user", which was a bit cruel. There is a mystery about the deadline, and Horacio's explanation is as good as any I've been able to come up with. But that - anonymous - Linux box is showing so many assorted errors and other problems that I wouldn't know where to start disentangling them.
	ID: 1260900 ·

alan Send message Joined: 18 Feb 00 Posts: 125 Credit: 311,600 RAC: 390	Message 1261563 - Posted: 17 Jul 2012, 15:31:56 UTC - in response to Message 1260900. Last modified: 17 Jul 2012, 15:38:44 UTC
	To me, "Aborted by user" means that the user deliberately aborted the task in BOINC Manager, and is a perfectly understandable message. "Not started by deadline" is mysterious - as you say, what deadline? - and as a replacement for "Aborted by user" makes no sense at all. The one is a deliberate action by the user, the other an automatic action by the BOINC client. The task hasn't reached any posted deadline, and if it had I would expect to see the "Timed out" message, even if the date had been deliberately set forward. It's an NVIDIA GPU client doing the reporting in any case, not Linux - my mistake. I've spotted the same substitution in another report, where I'd recorded the client response as "aborted" previously, and again it's an NVIDIA GPU client. My theory? It's been generated by the hallowed Hitchhikers method of pulling Scrabble letters out of a bag to be different from any other error message and to give us something non-threatening to argue about . . . :) ____________
	ID: 1261563 ·

Richard Haselgrove Volunteer tester Send message Joined: 4 Jul 99 Posts: 7459 Credit: 39,674,175 RAC: 33,868	Message 1261571 - Posted: 17 Jul 2012, 15:45:32 UTC - in response to Message 1261563.
	To me, "Aborted by user" means that the user deliberately aborted the task in BOINC Manager, and is a perfectly understandable message. "Not started by deadline" is mysterious - as you say, what deadline? - and as a replacement for "Aborted by user" makes no sense at all. The one is a deliberate action by the user, the other an automatic action by the BOINC client. That's precisely why I asked for the change. Previously, both cases got the "Aborted by user" message: now we should be able to distinguish between them. We actually had a case at Einstein where an experienced project adviser was confused by those inappropriate "Aborted" messages. The task hasn't reached any posted deadline, and if it had I would expect to see the "Timed out" message, even if the date had been deliberately set forward. My theory? It's been generated by the hallowed Hitchhikers method of pulling Scrabble letters out of a bag to be different from any other error message and to give us something non-threatening to argue about . . . :) But I'll have a ferret through the random message generator after maintenance and see what else drops out.
	ID: 1261571 ·

Message boards : Number crunching : Not started by deadline - canceled ???