Maybe no staff around this weekend to load more work to the FGRPB1G splitters? I believe I read somewhere else in another forum thread that the percent done is not a reflection on the actual amount of work left. That is in a post worrying about the project status page. And someone, Bernd? posted that the data coming off the sky surveys is continuously being added to the splitters.
Any thoughts on why GPU work is not being available?
There are a finite number of tasks associated with a given data file. The current file was LATeah1046L.dat which was first issued around late 27th Feb. These files tend to 'last' around 4 days which makes a new file pretty much 'due' right around now.
My guess is that all primary tasks for LATeah1046L.dat have been issued and whatever procedure is in place to transition to the next data file simply hasn't happened - for whatever reason.
I'd be extremely surprised if it's anything to do with completely running out of data. As Keith mentions, new data from the large area telescope (LAT) on board the Fermi satellite continues to be available so that the stats on the server status page to do with availability of work for this particular search can never reflect the true state of affairs. As they say, "It ain't over till the fat lady sings", and I haven't heard a peep out of her yet :-).
Worst case scenario - we might have to wait until Monday for this to be fixed. Best case scenario - new work might be stuck in some pipeline and some kind soul might intervene to clear the blockage. There hasn't been any sign of abnormal usage that I know of and the Staff do try to make sure that available work will outlast the weekend so maybe this could get fixed relatively quickly.
With multiple BOINC projects down at the same time, the probability that they all have a common cause is extremely high. Since the only thing all projects have in common is BOINC, the problem very likely comes from the BOINC system itself. The message I believe provides the best clue is in the message I am getting from SETI. At the end of each attempt to contact the server I get the following message:
3/3/2019 0:32:18 | SETI@home | [error] No scheduler URLs found in master file
I have seen this message before. It was a number of years ago, so I don't remember all the details, but I do believe it refers to a master list of send and receive addresses that BOINC uses to route all communications to and from each project. As I recall, the BOINC client software refreshes this from the BOINC host periodically. I believe it is once every 24 hours. If the BOINC host does not respond, the BOINC client, under its current logic, does not proceed until it can get the new list. It then sets a timer before trying again.
This would explain why, one by one, multiple projects are going down. It also seems plausible because the BOINC project, servers and software are basically run by the SETI people at Berkeley. SETI was the first to go down. It is most likely, therefore, that a problem at Berkeley, whether hardware or software, is affecting both the BOINC and SETI servers.
Unless someone like Kittyman or one of the other people with close connections with the Berkeley staff are lucky enough to get a quick answer from them during this major disaster, I guess we will have to sit back and wait until things get back up before we learn the whole story.
If anyone has more input into this theory, please let us know as soon as possible.
I think you’ll find the master url is per project. That way each project can manage their own. There is a url used to check connectivity when comms fail, it defaults to google.com. You can disable it or use a different URL via cc_config.
There is a separate project list (all projects) which is maintained at boinc.berkeley.edu, but not being able to download it would just mean it can’t update until next time. That is the BOINC server, not the Seti servers. The BOINC message boards are still working, suggesting there is no issues with their server.
With multiple BOINC projects down at the same time, the probability that they all have a common cause is extremely high. Since the only thing all projects have in common is BOINC, the problem very likely comes from the BOINC system itself. The message I believe provides the best clue is in the message I am getting from SETI. At the end of each attempt to contact the server I get the following message:
3/3/2019 0:32:18 | SETI@home | [error] No scheduler URLs found in master file
I have seen this message before. It was a number of years ago, so I don't remember all the details, but I do believe it refers to a master list of send and receive addresses that BOINC uses to route all communications to and from each project. As I recall, the BOINC client software refreshes this from the BOINC host periodically. I believe it is once every 24 hours. If the BOINC host does not respond, the BOINC client, under its current logic, does not proceed until it can get the new list. It then sets a timer before trying again.
This would explain why, one by one, multiple projects are going down. It also seems plausible because the BOINC project, servers and software are basically run by the SETI people at Berkeley. SETI was the first to go down. It is most likely, therefore, that a problem at Berkeley, whether hardware or software, is affecting both the BOINC and SETI servers.
Unless someone like Kittyman or one of the other people with close connections with the Berkeley staff are lucky enough to get a quick answer from them during this major disaster, I guess we will have to sit back and wait until things get back up before we learn the whole story.
If anyone has more input into this theory, please let us know as soon as possible.
Thanks!
Primegrid has tons of gpu work available, as does Collatz, Amicable Numbers, GpuGrid and Moo Wrapper.
IF you choose GpuGrid keep the cache very low as they do give bonus credits if you return the units within a pretty short deadline, it's explained on the website, and they ONLY accept higher end gpu's.
Maybe no staff around this
)
Maybe no staff around this weekend to load more work to the FGRPB1G splitters? I believe I read somewhere else in another forum thread that the percent done is not a reflection on the actual amount of work left. That is in a post worrying about the project status page. And someone, Bernd? posted that the data coming off the sky surveys is continuously being added to the splitters.
Betreger wrote:Any thoughts
)
There are a finite number of tasks associated with a given data file. The current file was LATeah1046L.dat which was first issued around late 27th Feb. These files tend to 'last' around 4 days which makes a new file pretty much 'due' right around now.
My guess is that all primary tasks for LATeah1046L.dat have been issued and whatever procedure is in place to transition to the next data file simply hasn't happened - for whatever reason.
I'd be extremely surprised if it's anything to do with completely running out of data. As Keith mentions, new data from the large area telescope (LAT) on board the Fermi satellite continues to be available so that the stats on the server status page to do with availability of work for this particular search can never reflect the true state of affairs. As they say, "It ain't over till the fat lady sings", and I haven't heard a peep out of her yet :-).
Worst case scenario - we might have to wait until Monday for this to be fixed. Best case scenario - new work might be stuck in some pipeline and some kind soul might intervene to clear the blockage. There hasn't been any sign of abnormal usage that I know of and the Staff do try to make sure that available work will outlast the weekend so maybe this could get fixed relatively quickly.
Cheers,
Gary.
I think a more interesting
)
I think a more interesting question, which I have not seen addressed recently, is what happens to O1OD1 in 22.9 days?
Gary I hope you are correct,
)
Gary I hope you are correct, with Seti being broken and this I had to attach to GPUGRID in order to keep busy.
MilkyWay is also down hard.
)
MilkyWay is also down hard. Crazy weekend if 3 major GPU projects are down at once.
With multiple BOINC projects
)
With multiple BOINC projects down at the same time, the probability that they all have a common cause is extremely high. Since the only thing all projects have in common is BOINC, the problem very likely comes from the BOINC system itself. The message I believe provides the best clue is in the message I am getting from SETI. At the end of each attempt to contact the server I get the following message:
3/3/2019 0:32:18 | SETI@home | [error] No scheduler URLs found in master file
I have seen this message before. It was a number of years ago, so I don't remember all the details, but I do believe it refers to a master list of send and receive addresses that BOINC uses to route all communications to and from each project. As I recall, the BOINC client software refreshes this from the BOINC host periodically. I believe it is once every 24 hours. If the BOINC host does not respond, the BOINC client, under its current logic, does not proceed until it can get the new list. It then sets a timer before trying again.
This would explain why, one by one, multiple projects are going down. It also seems plausible because the BOINC project, servers and software are basically run by the SETI people at Berkeley. SETI was the first to go down. It is most likely, therefore, that a problem at Berkeley, whether hardware or software, is affecting both the BOINC and SETI servers.
Unless someone like Kittyman or one of the other people with close connections with the Berkeley staff are lucky enough to get a quick answer from them during this major disaster, I guess we will have to sit back and wait until things get back up before we learn the whole story.
If anyone has more input into this theory, please let us know as soon as possible.
Thanks!
I think you’ll find the
)
I think you’ll find the master url is per project. That way each project can manage their own. There is a url used to check connectivity when comms fail, it defaults to google.com. You can disable it or use a different URL via cc_config.
There is a separate project list (all projects) which is maintained at boinc.berkeley.edu, but not being able to download it would just mean it can’t update until next time. That is the BOINC server, not the Seti servers. The BOINC message boards are still working, suggesting there is no issues with their server.
BOINC blog
Also, MilkyWay has been
)
Also, MilkyWay has been falling apart for weeks. I think they have hit rock bottom. Most BOINC projects (including GPUGrid) are still up.
Cherokee150 wrote:With
)
Primegrid has tons of gpu work available, as does Collatz, Amicable Numbers, GpuGrid and Moo Wrapper.
Primegrid: http://www.primegrid.com/
Collatz: https://boinc.thesonntags.com/collatz/
Collatz units pay the most credits of any gpu project if you use the optimization codes listed in the Number Crunching forum
Amicable Numbers: https://sech.me/boinc/Amicable/
GpuGrid: http://www.gpugrid.net/
IF you choose GpuGrid keep the cache very low as they do give bonus credits if you return the units within a pretty short deadline, it's explained on the website, and they ONLY accept higher end gpu's.
MoowWrapper: https://moowrap.net/
MW: Server is down hard and
)
MW: Server is down hard and site is not responding. Site was always available before when the db was not accessible.
SETI: 2nd Download server hasn't been working right since Tuesday maint. Site is back up today with DLing improvements
E@H: No GPU work.
Not related at all.