Index  | Recent Threads  | Unanswered Threads  | Who's Active  | Guidelines  | Search
 

Quick Go »
No member browsing this thread
Thread Status: Active
Total posts in this thread: 66
Posts: 66   Pages: 7   [ Previous Page | 1 2 3 4 5 6 7 ]
[ Jump to Last Post ]
Post new Thread
Author
Previous Thread This topic has been viewed 426095 times and has 65 replies Next Thread
adriverhoef
Master Cruncher
The Netherlands
Joined: Apr 3, 2009
Post Count: 2346
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: 2023-04-26 Update (system errors update)

Using ARP1_33870_119 as an example, 2 copies had been returned before the deadline. The third copy did not reply.

Mike, you're wrong.
The deadline for copies _0, _1 and _2 was 36 hours after release, as you can see by having a good look at the 'Due time' of copy _0 and the 'Sent time' of copies _3 and _4:

Result name        OS type   Status   Sent time           Due / Return time
ARP1_0033870_119_0 MSWin 10 No Reply 2023-05-07 00:15:33 2023-05-08 12:15:33
ARP1_0033870_119_1 MSWin 10 P.Val. 2023-05-07 00:15:35 2023-05-08 10:35:53
ARP1_0033870_119_2 MSWin S16 P.Val. 2023-05-07 00:15:57 2023-05-08 23:17:26
ARP1_0033870_119_3 MSWin 11 P.Val. 2023-05-08 12:15:40 2023-05-09 08:04:52
ARP1_0033870_119_4 MSWin 10 P.Val. 2023-05-08 12:16:05 2023-05-09 03:11:36


You can't change 36 hours into 72 hours just by thinking aloud "it is 72 hours, not 36 hours". laughing
So, only one copy - not 2 - made it before the deadline of 2023-05-08 12:15.

In succession, as soon as one copy (_0) didn't reply before the deadline, another two copies were sent (to sister and brother _3 and _4). At least those other two made it before their deadlines.

Also, the initial copies were sent out with a 36 hour deadline
confused
OK, so you've noticed. smile In this way it's all very confusing if you don't get it right the first time. Sorry.

Adri
----------------------------------------
[Edit 1 times, last edit by adriverhoef at May 9, 2023 11:44:50 PM]
[May 9, 2023 11:42:29 PM]   Link   Report threatening or abusive post: please login first  Go to top 
alanb1951
Veteran Cruncher
Joined: Jan 20, 2006
Post Count: 1317
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: 2023-04-26 Update (system errors update)

Mike,

I agree about "crunching something else more productively" :-) but it would need the restoration of the [pre-migration] grace days to stop the server prematurely re-issuing work for tasks that started pre-deadline but are still in progress[1]; I think we're stuck with this for now.

As for the reduced deadlines: if "reliable" (i.e. fast turnaround, and not prone to errors) systems get those 36-hour deadline tasks and don't have excessive buffers there's no reason (apart from very long delays in download or upload) that they should be unable to return them within 36 hours. Even the step-size 18 task(s) shouldn't take a modern machine more than 24 hours :-)

Adri,

I looked at your analysis in the other thread, and [looking at sent and returned times] it actually appears that the _4 task was triggered by the _2 task going No Reply (before it returned post-deadline). The point about unavoidable retries was valid either way :-)

Cheers - Al.

[1] The above said, I think the grace period mechanism is more important for tasks with longer initial deadlines; telling the clients the deadlines are shorter might produce a lot more "Not Started by Deadline" errors for tasks with a 5 day deadline (or longer) and perhaps encourage some folks to shrink their buffers somewhat, releasing more work for others :-) Indeed, specifically for ARP1 perhaps a 4-day deadline (with one day's grace) to try to reduce the number of units taking over a week to complete!

[Edited to remove a (longer) section rendered redundant by adriverhoef's post...]
----------------------------------------
[Edit 2 times, last edit by alanb1951 at May 10, 2023 12:29:17 AM]
[May 9, 2023 11:57:31 PM]   Link   Report threatening or abusive post: please login first  Go to top 
Mike.Gibson
Ace Cruncher
England
Joined: Aug 23, 2007
Post Count: 12594
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: 2023-04-26 Update (system errors update)

Sorry, I misread the return time for _2. adri's format is much clearer. However, with a quorum of 2 and 1 already returned it is less necessary to send the second resend. 1 would probably have been enough and in the end, neither were needed.

Mike
[May 10, 2023 2:23:29 AM]   Link   Report threatening or abusive post: please login first  Go to top 
adriverhoef
Master Cruncher
The Netherlands
Joined: Apr 3, 2009
Post Count: 2346
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: 2023-04-26 Update (system errors update)

Al:
I looked at your analysis in the other thread, and [looking at sent and returned times] it actually appears that the _4 task was triggered by the _2 task going No Reply (before it returned post-deadline).

d oh
You're right!
I already thought to myself that it was a bit strange, that missing _0's deadline generated 2 re-sends. And this is why, this explains it. So, _0 missing its deadline caused the release of copy _3, and _2 missing its deadline caused the release of copy _4.
Thanks for setting it straight!

Mike,
I also agree with you on "Those machines could have been crunching something else more productively", so you know. Anyhow, I'm afraid it is what it is. blushing

Adri
[May 10, 2023 7:30:33 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Hans Sveen
Veteran Cruncher
Norge
Joined: Feb 18, 2008
Post Count: 983
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: 2023-04-26 Update (system errors update)

Hi!
No wonder why an ordinary layman cruncher like me got confused!!

See my post https://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,41910

https://www.worldcommunitygrid.org/forums/wcg/viewpostinthread?post=685958

Edited to make it clearer, Thank You Adri!!
----------------------------------------
[Edit 1 times, last edit by Hans Sveen at May 12, 2023 4:56:05 PM]
[May 10, 2023 8:39:23 AM]   Link   Report threatening or abusive post: please login first  Go to top 
adriverhoef
Master Cruncher
The Netherlands
Joined: Apr 3, 2009
Post Count: 2346
Status: Offline
Project Badges:
Reply to this Post  Reply with Quote 
Re: 2023-04-26 Update (system errors update)

Hi!
No wonder why an ordinary layman cruncher like me got confused!!

See my post https://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,41910

Heya Hans,
Maybe this is some kind of help. Sorry if you already knew this.
Not only can you refer to a thread, you can also refer to a certain post. You can presumably handle this by clicking with your right mouse button (context), in the original post that you want to refer to (possibly open in another window or browser tab), on the 'chain' symbol in the lower right corner of that post.
The chain symbol is placed to the left of an 'unhappy face' and an 'arrow up' (in that same lower right corner) in the designated post.
When the context menu opens, you can find something like 'copy link' in it. When you choose 'copy link', the link to the post that you want to refer to is placed in your copy buffer (just like you can put the link to a thread in your copy buffer). In this way you can use the contents of your copy buffer to put it into an article that you want to post. blushing

Adri
[May 10, 2023 9:32:38 AM]   Link   Report threatening or abusive post: please login first  Go to top 
Posts: 66   Pages: 7   [ Previous Page | 1 2 3 4 5 6 7 ]
[ Jump to Last Post ]
Post new Thread