Missing credit server 129.32.209.201

[bruce]

When a problem like this is discovered, the timetable goes something like this.

Diagnose the problem on the server.
Correct the server problem for WUs uploaded after <date>
Scan old server logs and locate older WUs which were not credited
Issue re-credits for those uncredited WU.

Steps 3 and 4 are more tedious and tend to be delayed once step 2 has been completed.

[jeromesson]

Hi,
I get WUs from them quasi systematically now, so no more points for me at all. Usually project project:16927 and all goes well except the points. Not complaining but I have 2 questions:

Is it possible that it's the only ones a mac user can get? I remember that in a not so distant past, I had a variety of WUs coming from different projects. Now it's not the case anymore.

2nd question: do they have problems keeping up only with crediting the poins (which is a minor problem after all), or do they have problems keeping up with using the result of our folding (in which case they should slow down) ?

Thanks a lot.

[comixgoddess]

An update for reference ~ Starting on January 18, my WUs uploaded to this server were being credited - a total of 13 of them from the 18th through the 25th . Unfortunately, starting yesterday, the credits are no longer appearing, again.

2021-01-26 05:28:26 - project:16927 run:18 clone:593 gen:96
2021-01-26 13:01:05 - project:16927 run:22 clone:430 gen:81
2021-01-26 18:37:58 - project:16927 run:5 clone:1144 gen:21
2021-01-26 23:59:02 - project:16927 run:21 clone:348 gen:53
2021-01-27 05:12:37 - project:16927 run:26 clone:1437 gen:11
2021-01-27 12:47:22 - project:16927 run:23 clone:943 gen:71

[Hopfgeist]

For what it's worth, to help with localising the problem, I have not been credited for the following work units (although the collection server gave a credit estimate):

Server 129.32.209.201: Project: 16927 (Run 12, Clone 1268, Gen 9)
Server 129.32.209.201: Project: 16927 (Run 4, Clone 798, Gen 69)
Server 129.32.209.201: Project: 16927 (Run 15, Clone 810, Gen 104)
Server 129.32.209.201: Project: 16927 (Run 21, Clone 1257, Gen 2)
Server 129.32.209.201: Project: 16927 (Run 8, Clone 463, Gen 69)
Server 129.32.209.201: Project: 16927 (Run 23, Clone 563, Gen 90)
Server 129.32.209.201: Project: 16927 (Run 24, Clone 349, Gen 76)
Server 129.32.209.201: Project: 16927 (Run 17, Clone 1231, Gen 15)

And perhaps a few others, but these were on my two fastest machines with 12 threads each.


No problem for me, I'm in it for contributing to science, not for the points ...


Cheers,
HG.

[Hopfgeist]

Is it possible that it's the only ones a mac user can get? I remember that in a not so distant past, I had a variety of WUs coming from different projects. Now it's not the case anymore.

That seems unlikely. work units are assigned according to availability (which depends on which projects are active) and the number of advertised CPU cores. My machines with higher CPU count (12 or 24, all running the Linux client) also mostly get work units from project 16927, where there used to be more variety, whereas the slower machines (1, 3 or 4 CPUs) get work units from other projects.

HG.

[Neil-B]

Is it possible that it's the only ones a mac user can get? I remember that in a not so distant past, I had a variety of WUs coming from different projects. Now it's not the case anymore.

That seems unlikely. work units are assigned according to availability (which depends on which projects are active) and the number of advertised CPU cores. My machines with higher CPU count (12 or 24, all running the Linux client) also mostly get work units from project 16927, where there used to be more variety, whereas the slower machines (1, 3 or 4 CPUs) get work units from other projects.

Bernd

Slightly off the topic of thread... but are you running beta flags ... this sometimes explains why some people still get wuz when others not running the flas struggle ... not a suggestion to the op to run the beta flag if not part of the beta team just exploring reasons for different experience ... there also appears to be a bit of a rash of connection issues that might also explain differences

[jeromesson]

Is it possible that it's the only ones a mac user can get? I remember that in a not so distant past, I had a variety of WUs coming from different projects. Now it's not the case anymore.

That seems unlikely. work units are assigned according to availability (which depends on which projects are active) and the number of advertised CPU cores. My machines with higher CPU count (12 or 24, all running the Linux client) also mostly get work units from project 16927, where there used to be more variety, whereas the slower machines (1, 3 or 4 CPUs) get work units from other projects.

Bernd

Slightly off the topic of thread... but are you running beta flags ... this sometimes explains why some people still get wuz when others not running the flas struggle ... not a suggestion to the op to run the beta flag if not part of the beta team just exploring reasons for different experience ... there also appears to be a bit of a rash of connection issues that might also explain differences

Hi, sorry to be a little off topic. Not sure what you mean by "beta flags". So I guess I'm not concerned.

[Hopfgeist]

Is it possible that it's the only ones a mac user can get? I remember that in a not so distant past, I had a variety of WUs coming from different projects. Now it's not the case anymore.

That seems unlikely. work units are assigned according to availability (which depends on which projects are active) and the number of advertised CPU cores. My machines with higher CPU count (12 or 24, all running the Linux client) also mostly get work units from project 16927, where there used to be more variety, whereas the slower machines (1, 3 or 4 CPUs) get work units from other projects.

Bernd

Slightly off the topic of thread... but are you running beta flags ... this sometimes explains why some people still get wuz when others not running the flas struggle ... not a suggestion to the op to run the beta flag if not part of the beta team just exploring reasons for different experience ... there also appears to be a bit of a rash of connection issues that might also explain differences

Nope, no beta.

HG.

[Neil-B]

Members of the beta team run flags which mean they get projects before general release - this can sometimes mean they have wus when others dont - was checking if HG was running them - but not so that isnt what is causing the differences.

[comixgoddess]

Is it possible that it's the only ones a mac user can get? I remember that in a not so distant past, I had a variety of WUs coming from different projects. Now it's not the case anymore.

That seems unlikely. work units are assigned according to availability (which depends on which projects are active) and the number of advertised CPU cores. My machines with higher CPU count (12 or 24, all running the Linux client) also mostly get work units from project 16927, where there used to be more variety, whereas the slower machines (1, 3 or 4 CPUs) get work units from other projects.

HG.

For reference, I am running a single CPU slot with 4 CPUs, and ten of my last 12 work units have been project 16927 from this server, for a total of 43 16927 projects so far this month.

[Neil-B]

It is possible that there just aren't that many different cpu projects running at the moment - or at least that the server hosting 16927 has a much higher capacity to serve wus than the servers running other projects.

[PantherX]

...2nd question: do they have problems keeping up only with crediting the poins (which is a minor problem after all), or do they have problems keeping up with using the result of our folding (in which case they should slow down) ?...

Welcome to the F@H Forum jeromesson,

The way it works is when a WU is uploaded to the Servers (WS-Work Server or CS-Collection Server), the WS will automatically generate the next WU. The information of when it was assigned and returned (which is used to calculate points) would be collected by the Stats systems from the WS either in real-time or periodically (depending on the configuration and load). Thus, any WU that is uploaded to the WS, will be scientifically valuable even though you may not see points for a while (until the underlying issue is identified, resolved and the backlog processed).

When results are uploaded to CS, the assumption is that the CS to WS sync happens in which case, the WUs are again scientifically valid. However, on occasion, the CS to WS communication might be down and the WS assumes that the WU is lost so would issue another copy once the timeout has been reached. When the connection is restored between the CS and WS, previously processed WUs would be credited along with any currently assigned ones as those are genuine assignments.

Hopefully, this provides more context to some of the answers above

[SilvioMartin]

Here is some more debugging matter. All WUs were assigned from 129.32.209.201, are from the same project 16927, got a WORK_ACK in the log files, but they were not credited. There were some credited runs of the same project between these. Running on a Mac using FAH version 7.6.13. I got these lines from a "grep" command. I saved the log files, just in case that they are needed for debugging. Just let me know in case they are needed.

Code: Select all

logs/log-20210125-070008.txt:19:44:00:WU01:FS00:0xa7:Project: 16927 (Run 9, Clone 443, Gen 90)
logs/log-20210125-070008.txt:07:47:59:WU01:FS00:0xa7:Project: 16927 (Run 12, Clone 1592, Gen 0)
logs/log-20210128-182321.txt:11:54:34:WU00:FS00:0xa7:Project: 16927 (Run 31, Clone 1251, Gen 2)
logs/log-20210129-162153.txt:08:41:06:WU01:FS00:0xa7:Project: 16927 (Run 9, Clone 805, Gen 79)
logs/log-20210131-104808.txt:17:26:44:WU00:FS00:0xa7:Project: 16927 (Run 12, Clone 1400, Gen 5)
logs/log-20210131-104808.txt:10:56:34:WU01:FS00:0xa7:Project: 16927 (Run 17, Clone 1393, Gen 7)
logs/log-20210201-063653.txt:10:48:44:WU00:FS00:0xa7:Project: 16927 (Run 14, Clone 1831, Gen 9)
log.txt:06:38:36:WU00:FS00:0xa7:Project: 16927 (Run 10, Clone 1429, Gen 21) (today, Feb 1st)

This is how I checked whether results are there:

https://apps.foldingathome.org/wu?p=169 ... 443&gen=90
https://apps.foldingathome.org/wu?p=169 ... 1592&gen=0
https://apps.foldingathome.org/wu?p=169 ... 1251&gen=2
https://apps.foldingathome.org/wu?p=169 ... 805&gen=79
https://apps.foldingathome.org/wu?p=169 ... 1400&gen=5
https://apps.foldingathome.org/wu?p=169 ... 1393&gen=7
https://apps.foldingathome.org/wu?p=169 ... 1831&gen=9
https://apps.foldingathome.org/wu?p=169 ... 1429&gen=2

Happy debugging

[Hopfgeist]

Thus, any WU that is uploaded to the WS, will be scientifically valuable

I think this is the most important information for most. Thank you for clarifying that again.

Keep folding!
HG.

[NuclearWinter_S]

Hello everyone,

I also have this issue, no work units or credits from the server 129.32.209.201 and of project 16927 were counted in the last days. It is good to hear, that they will be still usefull for the science but will they be counted some day? Personally for me it is a little bit frustrating to spent the most expensive kWhs in Europe and probably in the world and not to see the result of the work