August Outage Aftermath
22 Aug 2009 0:15:51 UTC
Topic 194504
(moderation:
I am curious if others have observed anything curious with E@h following the "August Outage". To me, all operation seems normal except that the number of "Credit Pending" WU's has ballooned from claimed value of 5,727 just before the outage up to 15,402 now 3 days after the service was restored. Anyone have ideas about why the increase was so dramatic? (Claimed credit increase of about 10,000 equates to about 30,000 CS actual granted credit.)
Spot checking suggests that the other cruncher (wingman?) just kinda went away and isn't crunching. Has there been a significant loss in the number of active crunchers?
Well, the credits will come when/if they come, I'm sure. In the meantime, I just get antsi watching and waiting and wondering.
Stan
August Outage Aftermath
)
the reply button acts like a quote reply button and the quote button is missing.
I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
RE: the reply button acts
)
I'm sure that is true, but I don't understand the relevance. Is this a difference from before the outage? Can you expand on it, please?
Stan
Stan
RE: RE: the reply button
)
before i had a quote but and a reply button, i see the difference b/c i have to delete the quoted text.
I recommend Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
Thanks, zpm! My first
)
Thanks, zpm!
My first thought was that it was a reply to some other post!
I wondered at the omission of a "quote" box ... it is unlike most other forums I visit. But my memory was not good enough to know if there had been a "quote" box in this forum before.
Odd that a difference like that would arise from a recovery ... makes one wonder about the integrity of some change control procedures! :(
I honestly expected some modest temporary increase the the amount of pending work, since my cache ran dry only one day before E@h came back to life. But enough of the pending WU's were reported before the outage that it makes me wonder if some number of folks gave up on E@h and removed it from their boxes.
During the latter days of the outage, my logs showed several attempts to upload completed results per minute, which is perhaps consistent with the programmed delays of around 3 hours between individual WU attempts ... lessee ... 500+ results awaiting upload divided by 200 min delay between tries is 2.5 tries per minute. Yup! I regret that I caused that much useless traffic beating on E@h's "front door", but the programmed delay is a BOINC / E@h decision, not mine. Apparently the useless traffic didn't affect the recovery effort. And, all of the work eventually got there without apparent error. :) I was, in fact, pleased at how quickly E@h server(s) dealt with the backlog of work!
=====
Stan
Stan
RE: before i had a quote
)
There has never been a quote button at Einstein.
Other BOINC projects, yes: it was added a year or two ago. But Einstein has always been ultra-cautious about upgrading, and a true follower of "If it ain't broke, don't fix it".
However, now that they are testing a CUDA app, it's a shame they haven't added at least the code to display coprocessor details (hardware, driver version) in host lists.
I might be dense but if one
)
I might be dense but if one would want to post a reply without quoting one could press the "post to thread button", Right?
But then you wouldn't get the
)
But then you wouldn't get the "in response to Message 98919" indicator in the header.
If other contributors from other time-zones have posted while you're asleep, you may want that to appear to try and keep a conversation going.
I said, RE: Odd that a
)
I said,
Based on Richard's information, Einstein does not run by the maxim, "If it ain't broke, fix it 'till it is." So I shall trust the integrity of their change control!
Stan
RE: But then you wouldn't
)
Hmmm, I've always overlooked that indicator. This is the first time I'm 'seeing' it :D
Some versions of BOINC won't
)
Some versions of BOINC won't contact a project for a week or two after a long outage, unless a manual "update" is done. That could explain why many crunchers seem to have stopped crunching.