for English-speaking visitors, Welcome to our forum! Налаштування

[nikelong]

How to register (with pictures) for English-speaking visitors



Follow this link to switsh Your forum's language to English :
https://distributed.org.ua/forum/index.php?...act=idx&cal_id=

Це повідомлення відредагував ReMMeR: Feb 26 2011, 22:51

[zombie67]

How long should tasks with the new app run? I reset my windows machines this AM, per instructions. And the new tasks have been running for more than 4 hours now. Is this okay? Or a problem?

[Death]

looks like a problem.

as Yuri mention, you'd check in taskmanager that no lodci* tasks left in memory before resetting the project. if they keep running, project can't update them.

[zombie67]

Okay. I will abort the tasks and set to no new work. I will allow new tasks once we are told the problem has been fixed.

[SLinCA-Yuri]

Okay. I will abort the tasks and set to no new work. I will allow new tasks once we are told the problem has been fixed.

Dear zombie67, Alexiz and Death, and all other SLinCA-users,
thanks for your feedback!

It seems to be that first versions of application 'lodci' were not configured properly (even after updating their binaries),
and it is not easy to clean user machine and client BOINC Manager from the initial settings of faulty versions
(the authors of project can do this manually on their own machines, but it is not appropriate for all users).
That is why we decided to eliminate conflicting versions 'lodci16v4' and 'lodci32v4',
despite some unfinished results will be lost (it is not big loss for us, but unstable work is much bigger headache for users).
Afterwhile we will reconfigure and restart them again.

All application requirements are listed at page and in the column of 'Application' in tab 'Tasks' of your BOINC Manager.
If the correspondent version of 'lodci*'-application run more than runtime mentioned there, then it could be faulty, because chronometry is linear now.
Besides, checkpointing should work also, please check this.

Thanks to 'Alexis Kravtchenko' and 'Sirchyk' for your detailed reports!
Thanks to 're_SET', 'pvtt', 'zombie67', and 'Death' for your care!

[SLinCA-Yuri]

Our project Desktop Grid (DG) is connected to global Service Grid (SG) now!
April 11, 2011, 06:28 GMT
We are glad to announce that SLinCA@Home became the part of a global distributed computation infrastructure at production level thanks to the successful collaboration in the frame of the EU funded DEGISCO project and the EDGeS 3G Bridge technology from MTA SZTAKI. With this step, our scientific applications at SLinCA@Home Desktop Grid are executed now not only on the volunteers' home computers equipped by BOINC, but also on servers and clusters of Service Grid sites operated by HPC and Grid centers world-wide, which joined the International Desktop Grid Federation and offer capacities to numerous Desktop Grids.

The specificity of the situation: as it is known, SG was designed for high performance computing (HPC) and Grid-centers around the world and not equipped with BOINC managers!
In simple words this very complex interaction looks like channelling the workunits from SLinCA@Home DG to the SG-network (as a virtual volunteer participant), where the workunits automatically reconfigures for Linux-machines and the rules of SG, and then run on free (idle) SG resources.
The detailed statistics of such connection to the global SG-network through SG EDGeS 3G Bridge can be viewed on the page of such virtual volunteer participant ---> 'EDGES user'.

[SLinCA-Yuri]

Temporary withdrawal of large (> 1GB for RAM) tasks!

During the last weekend it has become clear that the big workunits (~ 2GB for RAM) in our BOINC-project were run in a very inefficient way:
- we obtained only 60 positive results from> 2000 tasks sent and they were from from Linux-machines only;
- they required large resources when the number of large workunits per machine was > 1 with blocking the RAM;
- they create bottleneck and do not allow small workunits to move to those volunteers who have less powerful machines.

Reasons?

Previous ones are as follows:
- incorrect temporary memory addressing for > 2GB in our clients for Windows,
- conflicts with system calls in Win 64-bit systems (this is fixed).
So we decided to cancel and return to them after more effective optimization and elimination of such bugs.

But we are grateful to those who managed to perform successfully these large workunits - they will be used by us!

Many thanks to re_SET, varador, Saenger, zombie67, DEATH, and many others who sent us their reports, observations and suggestions!
These conclusions and corrections have become possible only due to your help!

[SLinCA-Yuri]

Today outline of materials science applications will be given in Vilnius (Lithuania) at EGI User Forum 2011.

The seventh IDGF tutorial is co-located with the EGI User Forum 2011.
Review will be given for our materials science applications used in G.V.Kurdyumov Institute for Metal Physics of the National Academy of Sciences of Ukraine and ported to distributed computing environment on the basis of Desktop Grids technology in SLinCA@Home project.
They include:
(1) Monte Carlo (MC) simulation of defects in metals;
(2) image and video processing by MATLAB;
(3) molecular dynamics simulation by LAMMPS for research of defect aggregation in metal nanocrystals.
Please, see the program and get your virtual visitor bag.

Cheers!

[SLinCA-Yuri]

The updated versions appeared and 2x-credits were added for the long faulty workunits from the experimental 'lodci12x128'-version.
April 18, 2011, 10:50 GMT
The new updated versions of 'lodci'-application appear permanently after the planned upgrades and exclusion of configuration errors. The last error was related with misprint in configuration file of 'lodci12x128'-version.
Thanks to 'Saenger' and 'varador' for their reports!
'lodci12x128' was stopped and it will be started after re-configuration.
All participants will get renumeration credits that are equivalent to running the faulty workunits.
<--- this Soviet-style ink stamp means: "it was paid"

The project changes every day, and errors will appear constantly...
Therefore it is important to find them sooner than they cause any damage or cause your computers work in idle mode.
It is worth to note that most errors (> 80%) were found firstly by volunteers (ie by you).
We try to be open and report honestly to community about any faults that can appear in our project.
Thank you for your support!

[SLinCA-Yuri]

SERVER STATUS: Double-credits were awarded to all users, who successfully finished our test tasks, but had not got credits.
April 22, 2011, 07:50 GMT
Yesterday all participants, who successfully have finished our test tasks, but had not got any credits for them (because of 'Client error' status during tests), obtained double-credits. They are 2x-proportional to 'Claimed credits' for these faulty test workunits + some bonuses for other attempts. SLinCA project is at alpha-stage: i.e. it changes every day, and errors will appear constantly... Therefore it is important for us to find and fix errors sooner, than they cause any damage or cause your computers work in idle mode. It is worth to note that most errors related with distributed computing (>80%) were found firstly by volunteers (i.e. by you). Thank you for your support!

[laurenu2]

SERVER STATUS: Double-credits were awarded to all users, who successfully finished our test tasks, but had not got credits.
April 22, 2011, 07:50 GMT
Yesterday all participants, who successfully have finished our test tasks, but had not got any credits for them (because of 'Client error' status during tests), obtained double-credits. They are 2x-proportional to 'Claimed credits' for these faulty test workunits + some bonuses for other attempts. SLinCA project is at alpha-stage: i.e. it changes every day, and errors will appear constantly... Therefore it is important for us to find and fix errors sooner, than they cause any damage or cause your computers work in idle mode. It is worth to note that most errors related with distributed computing (>80%) were found firstly by volunteers (i.e. by you). Thank you for your support!

I completed about 1500 WU's all with 'Client error' I did not get any points at all

http://dg.imp.kiev.ua/slinca/results.php?u...names=0&state=5

Most that I seen Finnish looked like they completed OK
Were they flagged bad on your end or by my clients here
All my clients are still working your WU's Should I turn them Off

[SLinCA-Yuri]

SERVER STATUS: Double-credits were awarded to all users, who successfully finished our test tasks, but had not got credits.
April 22, 2011, 07:50 GMT
Yesterday all participants, who successfully have finished our test tasks, but had not got any credits for them (because of 'Client error' status during tests), obtained double-credits. They are 2x-proportional to 'Claimed credits' for these faulty test workunits + some bonuses for other attempts. SLinCA project is at alpha-stage: i.e. it changes every day, and errors will appear constantly... Therefore it is important for us to find and fix errors sooner, than they cause any damage or cause your computers work in idle mode. It is worth to note that most errors related with distributed computing (>80%) were found firstly by volunteers (i.e. by you). Thank you for your support!

I completed about 1500 WU's all with 'Client error' I did not get any points at all

http://dg.imp.kiev.ua/slinca/results.php?u...names=0&state=5

Most that I seen Finnish looked like they completed OK
Were they flagged bad on your end or by my clients here
All my clients are still working your WU's Should I turn them Off

You are right, I'll fix it in 2-3 days.
In addition, the new invalid tasks were not accounted at this stage of re0calculation of credits.

As far as your tasks have problem with 'Could not open ZIP_local file',
please, take into account the following information from official BOINC forum.
Maybe it will help to eliminate this problem?

Thanks again!

[SLinCA-Yuri]

As to the problem with 'Could not open ZIP_local file'

Many users with Windows machines (XP, Vista, 7, ...) suffer from the problem,
when their workunits were successfully calculated and credits were claimed,
BUT results were not packed in archive, because of some system limits on ZIPping files in BOINC-folder.
As a result, message 'Could not open ZIP_local file' was issued and 'Client state' becomes 'Compute error'
and 'Outcome' becomes 'Client error'.

To eliminate this problem, please, take into account the following information from the official BOINC forum.

We had Windows-machines running BOINC account with admin rights - so we did not notice this restriction.

Yesterday we tested this recommendation on Vista,
i.e. we imitated restricted (non-admin) rights and attempted to perform the workunits -
they generated this error.

Maybe it will help you to eliminate this problem?
We would appreciate very much, if you send us your feedback!
Many thanks to 'x3mEn' for his help!

[SLinCA-Yuri]

SERVER STATUS: Server will be offline during this weekend due to the planned repair of our regional power station.
May 13, 2011, 09:15 GMT
Tonight 'dark' Friday, 13-th day of May will be 'absolutely black' after 17:00 GMT and during weekend, because of the planned repair of our regional power station. That is why our server will be shut down up to May, 16th 05:00 GMT. Of course, you could crunch during this time in offline regime, if you set the BIGGER value for 'Additional work buffer' parameter (>1 day) in 'network usage' panel in 'Preferences' window in your BOINC Manager.
We apologize for the inconvenience and thank you for your support!

[SLinCA-Yuri]

SERVER STATUS: Server will be offline during this weekend due to the planned repair of our regional power station.
May 27, 2011, 15:o5 GMT
Tonight after 24:00 GMT and during weekend our server will be offline, because of the planned repair of our regional power station. That is why our server will be shut down up to May, 30th 05:00 GMT. Of course, you could crunch during this time in offline regime, if you set the BIGGER value for 'Additional work buffer' parameter (>1 day) in 'network usage' panel in 'Preferences' window in your BOINC Manager.
We apologize for the inconvenience and thank you for your support!

[zombie67]

I tried to crunch some tasks, hoping the problems have been resolved. But it appears that there are still problems. All 7 of 7 tasks errored out after 12.5 hours of run time. Please let us know when this is fixed, so I can return to the project.

http://dg.imp.kiev.ua/slinca/results.php?hostid=251

Here is a sample of the error:

stderr out
<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
Incorrect function. (0x1) - exit code 1 (0x1)
</message>
<stderr_txt>
Unrecognized XML in parse_init_data_file: hostid
Skipping: 251
Skipping: /hostid
Unrecognized XML in parse_init_data_file: starting_elapsed_time
Skipping: 0.000000
Skipping: /starting_elapsed_time
Unrecognized XML in parse_init_data_file: computation_deadline
Skipping: 1309229958.136000
Skipping: /computation_deadline
Could not open ZIP_local file.

</stderr_txt>
]]>