Troubles again at E@H?
19 Apr 2009 17:12:08 UTC
Topic 194300
(moderation:
I have several WU’s ready and the computers are waiting for new ones, but the database of E@H is not responding. The sites are also very very slow. Has anyone the same experience?
19/04/2009 18:52:10|Einstein@Home|Message from server: Server can't open database; is showing all day. After the problems of yesterday where solved I started my computers again. All of them because in the weekend I got power at a lower rate. However yesterday evening a new problem occurred at E@H. And again now information about this issue at the main site. Only a maintenance report at times. At this time mine computers a switched off (doing a bit of green crunching) and I am disappointed about all the problems in the last weeks. I think that everyone in Wisconsin is working hard to solve the problems, but at this time it is no fun to me.
Is it perhaps an idea to stop sending new WU’s to anyone and solve all the problems and test it for a week (especially over the weekend) and then let the community start crunching again?
Greetings from
TJ
Troubles again at E@H?
)
I believe it is up and running again now. Bernd said the database load problems, mentioned on the homepage, are still ongoing.
RE: I believe it is up and
)
According to the Server Status page, the Einstein scheduler has been turned off - perhaps to allow the transitioner and validation backlogs, also shown on that page, to clear. [Though the status page itself hasn't updated for two and a half hours, which is ominous].
Until the scheduler is back on, we won't be able to report any completed tasks or request any new ones.
Still no new tasks, does
)
Still no new tasks, does anyone have an idea when E@H will be up and running?
RE: RE: I believe it is
)
You are correct and it seems I misinterpreted what I read.
RE: Still no new tasks,
)
A new attempt to re-enable the scheduler is planned for Tuesday morning (German time).
CU
Bikeman
RE: RE: Still no new
)
Regarding the most recent status message:
"The database is still purging, at a rate of about 400,000 workunits per day. . ."
I've heard this story before--from Seti@Home, which even before transition to the BOINC environment, became a victim of its own success. Currently at just under a million users, 2.3 million host computers. S@H routinely goes offline EVERY Tuesday morning (California time) for ~6 hours primarily to purge & resynchronize the databases, as well as implement hardware upgrades, clean & reorganize the server closet, etc. etc. S@H's server farm is probably bigger than any other project (except IBM's World Community Grid), and a planned weekly outage seems only rational.
As (currently) the 2nd or 3rd biggest project host in the BOINC universe (depending on what metric you use), it seems equally rational for E@H to plan a weekly (or maybe semi-monthly) outage for similar purposes. Food for thought.
Best wishes :)