Posts by Ray Murray

1) Message boards : Ciencia : Dual Account (Message 577)
Posted 2 Oct 2020 by Ray Murray
Post:
Hi Pete
Log in with your original account,
On your account page, near the top, Change - email address
Change it to your new address
If that doesn't work because it says the address is taken by your NEW account, change the New account to a third address, go back to your original account and change it to the address that you want then abandon the new account as it will not have done any work,

Hope that works for you.
2) Message boards : Cafe : Queja - Sugerencias. (Message 565)
Posted 20 Aug 2020 by Ray Murray
Post:
And how hard would it have been for "The Ibercivis Team" to have posted that very same message in the News section? Aren't WE part of that team, too?
3) Message boards : Ciencia : No workunits (Message 562)
Posted 18 Aug 2020 by Ray Murray
Post:
Bummer still no work available and no response from the project admin in Last 4 days.
Last Admin post was 15th May with a faulty batch of WUs issued, a server outage and a couple of work famines but no explanation given.
They are hopefully (rightly) concentrating on analysis of the returned work but an update of that progress now and then would be nice.
I am still attached and eagerly await the next batch of work but others will not be as patient and will drift away if an occasional update or note of encouragement is not forthcoming.
4) Message boards : Ciencia : checkpoining (Message 555)
Posted 6 Aug 2020 by Ray Murray
Post:
Elapsed time regarding Checkpoints still isn't properly being reported to Boinc.
Generally I allow tasks to run to completion uninterrupted but today I had to restart my host to complete Windows and Antivirus updates. 1 task was at 69% after 6 hours or thereabouts and another at 20% after 2hrs. After restart, both continued from their % progress but the Elapsed Time reset to zero so instead of the full runtime, the reported time and hence time credited is only the 21/2 hours from restart to completion, so instead of c.200-300 credits, I only got 88.
While I realise the credits themselves are worthless and the Science priceless, others may find the discrepancy unacceptable. I imagine there is only a single line missing from the Tasks template. If I knew where, I would point to it. Other crunchers may have a better idea where to look. If I find it, after some research tomorrow, I'll report back.

[Update after less research than expected]
Found it near the bottom of this page:
https://boinc.berkeley.edu/trac/wiki/BasicApi

Timing information
int boinc_wu_cpu_time(double &cpu_time);

gets the total CPU time (from the beginning of the work unit, not just since the last restart).

double boinc_elapsed_time();

returns the elapsed runtime (i.e. wall-clock time during which the job was not suspended) since the start of the current episode. The elapsed time from earlier episodes is in APP_INIT_DATA::starting_elapsed_time (only from 6.10+ clients).
5) Message boards : Ciencia : Jobs crashing. (Message 554)
Posted 6 Aug 2020 by Ray Murray
Post:
Welcome, Emma
Your host has returned 2 completed, validated and credited tasks so at first glance, it would seem to be working.
Tasks take anything from 3 to 10 hours to complete depending on what their initial internals settings are. There is no way to guess that at the outset. Progress % as reported by Boinc is not linear. It may start off looking that way but once it gets to 10%, it will then not rise incrementally until there is a jump to 20% and so on. I have also seen 69% and 84% so the jumps might depend on whereabouts any phases of the computations complete.
There was previous discussion on whether they checkpoint. That might be at the % jumps but I haven't tested that as I have Boinc set up to let ANY task from ANY project to run to completion without being swapped out. [Switch between tasks every 1500 minutes] If there is continual switching, some progress may be lost and rewound unless "Leave in memory while suspended" is ticked.

Hope that helps.
6) Message boards : Ciencia : Jobs crashing. (Message 542)
Posted 22 Jul 2020 by Ray Murray
Post:
Good, I'll get some as my other Projects finish their running tasks.
I did notice this morning that the site was totally unreachable so there might have been some other issue going on, presumably now fixed.
It would have been nice to see something from the Project as a News item or a Boinc message such as "It was broken. Our bad. We've fixed it, now."
7) Message boards : Ciencia : Jobs crashing. (Message 531)
Posted 18 Jul 2020 by Ray Murray
Post:
All the good WUs seem to gone and we're back to the faulty R60000s that got sent to the back of the queue as I've just rattled through 160 or more in succession until I got bored manually hitting the update button. Many of them have been the 7th resend and are getting bumped as having had Too many Errors so at least they won't be going back into the queue, but a lot will have to be sent out again, and again unless there is intervention from the Project.
I had hoped that the Admins would have taken the time while we had some good work this week to notice that there was something broken with these R60000s and rectified the problem, but alas, no.
[I'd temporarily upped my cache to see how many I could catch so that they can be culled and replaced with WUs that work but now I've exceeded my quota of tasks for the day so won't be able to capture any more until tomorrow]

Rosetta has lost a bunch of contributors recently (I think) due to a lack of communication.
I fear the same may happen here as there have been a couple of job shortages and now this faulty WU problem but no information regarding any of that from the Project. They may, rightly, be concentrating on the science but the Projects that hold on to their contributors longer, and therefore get more work done by them, are the Projects that communicate regularly with updates on progress or problems.
8) Message boards : Ciencia : Jobs crashing. (Message 529)
Posted 15 Jul 2020 by Ray Murray
Post:
When I looked earlier there were 60,000+ unsent tasks in the queue but I looked again a few mins ago and saw 120,00+. Thinking this might be a whole new batch, I allowed 3 to download. 1 was a 5th resend of the known-to-be-faulty R60000s and failed as expected, but the other 2 were R80000s and the first of those is chugging along fine after 15 mins.
I suspect, however, that we will have to work our way through multiple resends of the bad R60000s as, having just got one, I have no reason to doubt that the 60,000+ of them I saw earlier are still lying in wait. If the Project Admin were to remove them all, or fix whatever is broken with them, that would probably be a better solution than simply waiting for them to fail the required number of times.
9) Message boards : Ciencia : Jobs crashing. (Message 520)
Posted 14 Jul 2020 by Ray Murray
Post:
Todays batch of 13072020_R60000_L000007 or ....8 are failing in download.

Update before posting:
2 have just downloaded successfully but being from yesterday's batch, have returned the Compute error.
Back to NNT. I'll check back later.
10) Message boards : Ciencia : Jobs crashing. (Message 516)
Posted 13 Jul 2020 by Ray Murray
Post:
Earlier I completed some 13072020_R40000_L000006 tasks successfully but ALL of the 13072020_R60000_L000009 flavour fail after only a few seconds. Of those that I have checked, all wingmen are encountering the same issue so just a bad batch, I guess. Setting No New Tasks overnight and will check back in the morning.
11) Questions and Answers : Unix/Linux : Too much 202 (0x000000CA) EXIT_ABORTED_BY_PROJECT errors (Message 482)
Posted 26 Jun 2020 by Ray Murray
Post:
Often the reason for a Project Aborting tasks at their end is that they may have found a flaw in a batch of tasks from which returned results would be useless so rather than wasting your time and theirs, they Abort the whole batch.
12) Message boards : Ciencia : Maximum task run time? (Message 450)
Posted 1 Jun 2020 by Ray Murray
Post:
Si, tienen checkpoint cada 10%, las que no lo tenían eran las de Abril
Yes, they have checkpoint every 10%, the ones that didn't have it were April's‎
I'm not convinced the rb03s and rb04s have checkpointing as I looked in the Properties of a few tasks and the cpu time and the cpu time since last checkpoint were the same, certainly beyond 20hrs so I let them run to completion without any harassment from me. Just to make sure, I've got "switch between tasks every" set to 1500 minutes so they don't get swapped out to let something else run, like a WCG or Rosetta (backing all the dogs in the fight, with Folding on another machine and Dreamlab on the phone)
13) Message boards : Ciencia : Aborted by project (Message 383)
Posted 20 May 2020 by Ray Murray
Post:
Usually this means that the Poject has uncovered a fault in the WUs whereby any result they returned would not be useful so rather than waste your time and theirs, they have decided to Abort them to make way for useful work.




©2021 CC-BY-4.0