I have been getting the following messages today from the project.
2/9/2007 9:13:35 AM|Einstein@Home|Message from server: Server can't open database
2/9/2007 9:13:35 AM|Einstein@Home|Project is down
I have noticed very slow load times for the web pages today also.
I'm sure everybody is seeing the same at the moment.
Approximately 15 hours ago, I installed the EAH project on a new Core 2 Duo laptop as a way of giving the machine a good stress test over the weekend until it goes to its new home (minus BOINC) early next week. The installation was fine and the initial downloading of approximately 20 or so results was fine, except that those results came from about 10 different large data files. Not only that, as work was crunched and more work requested, at least six of those large data files were marked for deletion after yeilding up only 1-3 results each.
It would appear to me that we are in a period of high server load like we were when the S5R1 run was in the final stages of completion, with lots of dregs of work being sent out. Why this should be is a bit of a puzzle since the server status says that this run has close to 80% still left to run. Maybe there are runs within runs or something like that.
Whatever it is, it's most frustrating to see such large quantities of data being downloaded, only to be thrown away shortly thereafter. Hopefully we will get some sort of improvement in this apparently quite inefficient work distribution scheme at some stage when the Devs get a chance to look at it.
I've commented previously in this thread on the high server load that appears to happen at times when work distribution seems to be sub-optimal.
Another possible reason for the extra load on the server is due to the fact that team Synergy has picked this as their project of the month. That means their whole team has concentrated their efforts on climbing the ranks here for this month. This caused some problems for uFuids last month too. It may not be such a good idea for them to be doing this in light of the problems this may cause the projects.
Our (BoincSynergyies) combined RAC is 70K, not really a huge load in comparison to some teams, but an added load?, certainly. However, when the POTM started we were at 51K, so we've only added 19K's worth of additional load.
Not to mention the project started acting "funky" in late January, while we were still doing Ufluids.
It looks like it'll take a while until they solve their "Large BOINC Database" Problems.
Maybe they'll be forced to shutdown generation and distribution of new work to give the Validator/Deleter Jobs the chance to catch up (SETI went through very similar problems several times).
In the meantime I just see Pending reaching new record-breaking levels every day; when the validator works it seems to get only ~70-80% done e.g. of my daily production.
Strangely, I still maintain a full suppliy of WorkUnits, no single indication of "no new Work" so far.
It looks like it'll take a while until they solve their "Large BOINC Database" Problems.
........snip....
Strangely, I still maintain a full suppliy of WorkUnits, no single indication of "no new Work" so far.
No new work here now. Got one lot after a messy looking business; then the project slammed shut last night!"
all my Win machines still seem to get work. However, My linux machine hasn't seen a wu in over a week. The last wu for it was reported 12 Feb, and it had a 3 day cache, so sometime around Feb 9 or 10 it just couldn't get work.
I cannot even get a new machine attached to the project. Been trying
for about 4 days. It has ago, then gives up saying ... server cannot open
database.
The project has been generating apparently spurious 'project down' responses -- this has been going on for a while now. Not a big deal when uploading data -- I find that within 20 seconds, the 'project down' is miraculously solved and the upload succeeds.
The problem with downloads is that if you encounter that 'project down' -- you can't succeed with a download right after that as you get a 'you tried to download work too recently' response.
The thing with Einstein is that there are apparently *multiple* issues going on. One is this connectivity thing (project down reports), another is the growing number of pendings. Also, they are going to bringing out a new batch and style of work units. Lastly, there are apparently, in addition to the software issues, some hardware problems needing resolution.
Unfortunately, there really is not that much communication going on from project administration folks. So folks get into speculative mode a fair amount. There are some people participating on the message boards that try to mitigate this by providing explanations based on what they believe is going on. But to a certain degree, in the absence of explanations on the home page of status issues, the good folks offering explanations here in the various different message boards and threads (they end up posting on multiple boards and multiple topics because there is no centralized status explanation being offered so folks post all over the place) are seeming to be more and more frustrated with the noise level generated by lowly folks such as us.
Quote:
I cannot even get a new machine attached to the project. Been trying
for about 4 days. It has ago, then gives up saying ... server cannot open
database.
RE: I have been getting the
)
I'm sure everybody is seeing the same at the moment.
Approximately 15 hours ago, I installed the EAH project on a new Core 2 Duo laptop as a way of giving the machine a good stress test over the weekend until it goes to its new home (minus BOINC) early next week. The installation was fine and the initial downloading of approximately 20 or so results was fine, except that those results came from about 10 different large data files. Not only that, as work was crunched and more work requested, at least six of those large data files were marked for deletion after yeilding up only 1-3 results each.
It would appear to me that we are in a period of high server load like we were when the S5R1 run was in the final stages of completion, with lots of dregs of work being sent out. Why this should be is a bit of a puzzle since the server status says that this run has close to 80% still left to run. Maybe there are runs within runs or something like that.
Whatever it is, it's most frustrating to see such large quantities of data being downloaded, only to be thrown away shortly thereafter. Hopefully we will get some sort of improvement in this apparently quite inefficient work distribution scheme at some stage when the Devs get a chance to look at it.
I've commented previously in this thread on the high server load that appears to happen at times when work distribution seems to be sub-optimal.
Cheers,
Gary.
[quote It would appear to me
)
[quote
It would appear to me that we are in a period of high server load like we were when the S5R1 run was in the final stage
I've commented previously in this thread on the high server load that appears to happen at times when work distribution seems to be sub-optimal.
Message from server: Server
)
Message from server: Server can't open database
Google indexing maybe? Those sick Google bots are often not much different from a DoS attack and robots.txt does not cover all dynamic pages.
Google is allowed on "create*", "forum*", "stats/*", "top_*", "view_", probably some more that I forgot now.
The robots from www.emeraldshield.com ignore robots.txt completely, it must be a fake anti-spam company.
Another possible reason for
)
Another possible reason for the extra load on the server is due to the fact that team Synergy has picked this as their project of the month. That means their whole team has concentrated their efforts on climbing the ranks here for this month. This caused some problems for uFuids last month too. It may not be such a good idea for them to be doing this in light of the problems this may cause the projects.
Steve
98SE XP2500+ @ 2.1 GHz Boinc v5.8.8
Our (BoincSynergyies)
)
Our (BoincSynergyies) combined RAC is 70K, not really a huge load in comparison to some teams, but an added load?, certainly. However, when the POTM started we were at 51K, so we've only added 19K's worth of additional load.
Not to mention the project started acting "funky" in late January, while we were still doing Ufluids.
It looks like it'll take a
)
It looks like it'll take a while until they solve their "Large BOINC Database" Problems.
Maybe they'll be forced to shutdown generation and distribution of new work to give the Validator/Deleter Jobs the chance to catch up (SETI went through very similar problems several times).
In the meantime I just see Pending reaching new record-breaking levels every day; when the validator works it seems to get only ~70-80% done e.g. of my daily production.
Strangely, I still maintain a full suppliy of WorkUnits, no single indication of "no new Work" so far.
RE: It looks like it'll
)
No new work here now. Got one lot after a messy looking business; then the project slammed shut last night!"
John
all my Win machines still
)
all my Win machines still seem to get work. However, My linux machine hasn't seen a wu in over a week. The last wu for it was reported 12 Feb, and it had a 3 day cache, so sometime around Feb 9 or 10 it just couldn't get work.
I cannot even get a new
)
I cannot even get a new machine attached to the project. Been trying
for about 4 days. It has ago, then gives up saying ... server cannot open
database.
Nairb
The project has been
)
The project has been generating apparently spurious 'project down' responses -- this has been going on for a while now. Not a big deal when uploading data -- I find that within 20 seconds, the 'project down' is miraculously solved and the upload succeeds.
The problem with downloads is that if you encounter that 'project down' -- you can't succeed with a download right after that as you get a 'you tried to download work too recently' response.
The thing with Einstein is that there are apparently *multiple* issues going on. One is this connectivity thing (project down reports), another is the growing number of pendings. Also, they are going to bringing out a new batch and style of work units. Lastly, there are apparently, in addition to the software issues, some hardware problems needing resolution.
Unfortunately, there really is not that much communication going on from project administration folks. So folks get into speculative mode a fair amount. There are some people participating on the message boards that try to mitigate this by providing explanations based on what they believe is going on. But to a certain degree, in the absence of explanations on the home page of status issues, the good folks offering explanations here in the various different message boards and threads (they end up posting on multiple boards and multiple topics because there is no centralized status explanation being offered so folks post all over the place) are seeming to be more and more frustrated with the noise level generated by lowly folks such as us.