gaia@home

Posts by PDW

1) Message boards : Number crunching : A lot of errors (Message 674)
Posted 3 hours ago by PDW
Post:
COLD REBOOT & PROJECT DETACH/REATTACH made no difference.
Still getting the -195 exit error ONLY on 4_Gaia@home tasks after a few seconds of trying to run.

Running Ubuntu 18.04.6 LTS [5.4.0-200-generic|libc 2.27]

The boinc_finish(195) is not the real error.
The "app exit status: 0x8b" is the problem.

Your host got this for a task:
05:03:10 (30950): wrapper (7.15.26016): starting
05:03:10 (30950): wrapper (7.15.26016): starting
05:03:10 (30950): wrapper: running ../../projects/gaiaathome.eu_gaiaathome/4_gaia@home[20241130.0949]_x86_64-pc-linux-gnu ()
05:03:11 (30950): 4_gaia@home[20241130.0949] exited; CPU time 0.002265
05:03:11 (30950): app exit status: 0x8b
05:03:11 (30950): called boinc_finish(195)

The next host to complete the task got this:
14:55:39 (1649670): wrapper (7.15.26016): starting
14:55:39 (1649670): wrapper (7.15.26016): starting
14:55:39 (1649670): wrapper: running ../../projects/150.254.66.104_gaiaathome/4_gaia@home[20241130.0949]_x86_64-pc-linux-gnu ()
rm: cannot remove 'model.obj': No such file or directory
rm: cannot remove 'observations.dat': No such file or directory
15:18:14 (1649670): 4_gaia@home[20241130.0949] exited; CPU time 1326.155024
15:18:14 (1649670): called boinc_finish(0)

These were the recorded memory values for the completed task:
Peak working set size 223.45 MB
Peak swap size 45.67 GB
Peak disk usage 0.22 MB

The Peak swap size is enormous and I've no idea why it is that value, it certainly can't be using all that Swap space at once as ALL my systems would die if that was real given the number of tasks being run simultaneously.

Possibly the 4_gaia tasks ask for more initial memory than the 6_gaia tasks do and your host is not able to supply that amount.
Just a guess, you would need zupa to explain what the code is trying to do and why your host is failing all the 4_gaia tasks.

I would expect if you were missing libs or higher versions then the log would show that like it does for libquadmath.

Also, no idea what or why the "rm: cannot remove" errors appear in the completed tasks stderr log.
2) Message boards : Number crunching : A lot of errors (Message 634)
Posted 5 days ago by PDW
Post:
Suddenly seeing lots of errors on the tasks 4_Gaia@home v1.00 x86_64-pc-linux-gnu
"Exit status 195 (0x000000C3) EXIT_CHILD_FAILED"
https://gaiaathome.eu/gaiaathome/results.php?userid=5503&offset=0&show_names=0&state=6&appid=278

And still getting same error message. Up to 41 failed/error now.

I'd suggest you regenerate !

Those tasks are fine on other hosts so it suggests it is a local problem, a project reset (losing current work) would get new executables.
3) Message boards : Number crunching : Gaia 6 is new (Message 585)
Posted 3 Nov 2024 by PDW
Post:
Thank you, mine have all been given credit :)
4) Message boards : Number crunching : Gaia 6 is new (Message 583)
Posted 3 Nov 2024 by PDW
Post:
You said "created on 3 Nov 2024, 0:07:00 UTC"

zupa said it was fixed at 3 Nov 2024, 12:11:39 UTC. That's for tasks assigned after that point.

He also said "I think, that the time of comeback of results will be important for this problem ...."
But yes, I think I was lucky and got a working one.
I have no evidence to prove that all the "bad" tasks are out of the system (not including the Q2 tasks already sent/ready to be sent).

If he is going to master giving back credit for Valid tasks that get 0 credit before they get removed from the database then there is no problem and those tasks can be worked through.
5) Message boards : Number crunching : Gaia 6 is new (Message 581)
Posted 3 Nov 2024 by PDW
Post:
That task completed, a bit longer than usual, but was valid and got credit. No Q2 task.

That earlier workunit I linked was completed as Valid and 0 credit for both of them.
That was after you said it should be fixed but it was already in a Q2 situation.
6) Message boards : Number crunching : Gaia 6 is new (Message 578)
Posted 3 Nov 2024 by PDW
Post:
Well in 3 hours time 6_12493_0 created on 3 Nov 2024, 0:07:00 UTC should be completed.
I can't tell if it has a hidden Q2 duplicate but you think uploading the completed WU in 3 hours time will result in it getting credit.
That would happen anyway if it does not have a Q2 duplicate.
The clock is ticking...
7) Message boards : Number crunching : Gaia 6 is new (Message 576)
Posted 3 Nov 2024 by PDW
Post:
Yes, those are Resends for failed tasks, 1 WU fails and it gets 1 resend to replace it.

Have downloaded what I hope are new tasks, need to find the running 6_G task in my account's task list to get details for creation time.

PS. ( I must learn how to do it ....) <-- always the sign of a good admin :)
8) Message boards : Number crunching : Gaia 6 is new (Message 572)
Posted 3 Nov 2024 by PDW
Post:
Good, hopefully !

Is this a fix for all existing tasks (4, 5 & 6) like the workunit I posted earlier or just for new work sent out since you think you fixed it ?

Tasks are all meant to be Quorum 1 by design, we should not see Quorum 2 on any new tasks created ?
9) Message boards : Number crunching : Gaia 6 is new (Message 569)
Posted 3 Nov 2024 by PDW
Post:
Last night, mine started at 6_###_0 and in general sequence.
I looked at a lot of them after the first result appeared as Q2 and they all said Q1 before a result was uploaded.
I was surprised that they had started running in preference to others that had been downloaded much earlier, I was in bed so didn't get up to check machines but deadlines looked to be 11/11 for them. I hoped it was because it was a new app and that mostly my 4s and 5s would get processed in date/time order.

Resends are getting much shorter deadlines, for example this one: https://gaiaathome.eu/gaiaathome/workunit.php?wuid=257563 shows a deadline of 5 Nov 2024, 10:38:07 UTC.
The tasks will get 0 credit if ever a second WU completes successfully.
10) Message boards : Number crunching : Gaia 6 is new (Message 567)
Posted 3 Nov 2024 by PDW
Post:
Give it time, you'll get there !

I've seen others with a host that has processed loads and not a single 0 credit, their other hosts have the 0 credits.
I haven't been able to spot what might be a common factor, it looks like it is at the server end.

I have never seen tasks sent out as that appear as Quorum 1 then magically turn into Quorum 2 when the first WU gets to the server, which it always was because two WUs were created.
I've seen details of a second WU be hidden but they always showed as Quorum 2 to begin with.
Why some become Q2 when they are all meant to be Q1 ?
I assume the Validator has a meltdown when it tries to validate that the work for both is correct, it is and they are marked Valid, but not when it has 2 WUs instead of just the one so they get 0 credit.
11) Message boards : Number crunching : Gaia 6 is new (Message 565)
Posted 3 Nov 2024 by PDW
Post:
I find tasks where the system sends 2 at the same time despite the settings that only one is to send (both successfully and zero loans) :(

Yes, lots of of them.
It is doing it at work generation time and not as a result of receiving a completed task and thinking I'll get a second opinion on that and generate a second task to compare it with when it completes.
12) Message boards : Number crunching : Gaia 6 is new (Message 563)
Posted 3 Nov 2024 by PDW
Post:
The ones that become Valid and get 200 credits stay as Quorum 1, those that become Quorum 2 get 0 credit.
This also seems to be the reason why other applications got 0 credit on occasion.
6 just seems to do it to a much higher proportion of the tasks.
13) Message boards : Number crunching : Gaia 6 is new (Message 562)
Posted 3 Nov 2024 by PDW
Post:
zupa, I do not know what you have done but 6 is bad for most people.
Some people are getting credit for some of their 6 tasks but a lot are not.

The problem seems to be that the tasks we receive look like they are Quorum 1 but when they get to pending you can then see that a second task was created at around the same time as the first task and they are now Quorum 2.

The ones that become Valid and get 200 credits stay as Quorum 1, those that become Quorum 2 get 0 credit.
14) Message boards : Number crunching : Gaia 6 is new (Message 561)
Posted 2 Nov 2024 by PDW
Post:
1st task is Valid and credit 0
2nd task is pending wingman
Remaining tasks in progress are quorum 1 !

Update: 2 Valid with credit 0
15) Message boards : Number crunching : 0 points for valdiated task ? (Message 551)
Posted 31 Oct 2024 by PDW
Post:
I did click the first link. Can't comment on your issues since you choose to hide your hosts. It looks like all the 0 credit issues have to do with Gaia_4 tasks which were faulty and likely cancelled in bulk by the admins to make room for the Gaia_5 tasks.

No. I only have seen it on 5_Gaia tasks, with only 2 tasks sent out for the workunits.
16) Questions and Answers : Web site : Cometary Science Badge not displaying (Message 434)
Posted 21 Oct 2024 by PDW
Post:
Don't get too excited Conan, they are just the standard BOINC bronze, silver and gold RAC badges despite their title.
The links point to the old http site with the numbers in the URL.
17) Message boards : Number crunching : 1004_gaia@home[20220808.1523]_x86_64-pc-linux-gnu executable (Message 338)
Posted 8 Aug 2022 by PDW
Post:
The above file doesn't have the execute bits set so they fail to run.
The 1004_Gaia one does.
18) Message boards : Number crunching : Points (Message 194)
Posted 18 Jan 2021 by PDW
Post:
When I ran it recently it worked fine on the hosts to begin with.
After a period of time some hosts started to get 1 or 2 results with 8.33 credit, after a while those hosts would not get good credit so I stopped them running the project. Other hosts continued to run fine without any problems, I suspect once whatever triggers them to go bad they stay bad, perhaps a reboot [I think I tried that without success though] or something else is needed to reset the behaviour. Hosts that have been bad in the past will start working again when I try months later.

I can see no pattern within the tasks that are sent out, a bad task will get resent out to another host and it will complete and get full credit.

What does concern me when they are bad is why everybody gets 8.33 points no matter how long it takes to complete.
Is 8.33 some default value within the validation procedure that is used ?
Is something within the returned results triggering the validation procedure to drop out and give the 8.33 credits ?
Normally tasks that take longer on the same host get more credit than shorter ones.

Just some observations :)
19) Questions and Answers : Getting started : Can't create an account or join (Message 75)
Posted 3 Oct 2020 by PDW
Post:
Anybody got a workaround for team creation?
One workaround would be to join another team ;)

It is a problem that requires the server code to be changed that Sergey has already linked to, it needs the admin (Zupa) to action.
20) Message boards : Number crunching : Why project do not export statistics? (Message 51)
Posted 25 Sep 2020 by PDW
Post:
Check with Willy, Bok has them loading at Free-DC.


Next 20

©2024 GAVIP-GC