1) Message boards : Number crunching : Error with one of the new WUs (Message 14799)
Posted 21 Sep 2006 by TPR_Mojo
Post:
Lots of disk space errors here too, even though LHC is only using about 17Mb of neraly 2Tb disk. A wee glitch methinks.
2) Message boards : Number crunching : Host corruption solved? (Message 11092)
Posted 28 Oct 2005 by TPR_Mojo
Post:
Chrulle, do you have any plans to "manually" fix any hosts that aren't getting caught automatically? As of the last XML export (10 hours ago), I see about 1,800 hosts that are still showing invalid data. Some of them are probably no longer active. Some people probably reset the project in hopes of fixing the problem and got a new host ID. Now they can't be merged because of the bad data. It would be nice to get these hosts out of the system since they are messing up some of my stats queries. If you have no plans to do anything about them I suppose I will find a workaround on my end.


Delete the lot - they are quite simple to identify.
1800 hosts that everyone no longer has to work around.
3) Message boards : Number crunching : Host corruption solved? (Message 11030)
Posted 26 Oct 2005 by TPR_Mojo
Post:
So far so good. I have several corrupted host records on my account, but the hosts they relate to have created new records for themselves and (crossed fingers) the corruption doesn't appear to have affected these new records.
4) Message boards : Number crunching : Host corruption (Message 10942)
Posted 25 Oct 2005 by TPR_Mojo
Post:

this seems to confirm my suspicion that the problem isn't in host-server-communication but somewhere between the validator and the database (see some of my earlier posts)
I already send a mail with what I have found to Chrulle.

Results are still being validated and credit is given to the correct hosts, so I guess we'll just have to wait and see what happens next.


Good job, I'm not too fussed really, only provided my dodgy hosts because Chrulle asked for them. Work is still being done, BOINC isn't crashing, so a wee bit of patience - hopefully they have enough information now to track the problem down.
5) Message boards : Number crunching : Host corruption (Message 10938)
Posted 25 Oct 2005 by TPR_Mojo
Post:

let me guess, you also got credit for that result?


Yup, must edit quicker, my bad :)
6) Message boards : Number crunching : Host corruption (Message 10936)
Posted 25 Oct 2005 by TPR_Mojo
Post:
Host 68191 just returned a result (4068306) at 18:09 UTC from BOINC 5.2.4 (Linux) which corrupted the host record.

Result is fine, was nominated canonical result and granted credit.

7) Message boards : Number crunching : Host corruption (Message 10924)
Posted 25 Oct 2005 by TPR_Mojo
Post:
Corrupted hosts on my account

37059
57498
34303
34351
57463
65799
57431
36703
24898
57462
37063
36752

All Windows machines were running BOINC 4.45, Linux 4.45 or 4.72.

ALL machines were upgraded to 5.2.2 (Windows) or 5.2.4 (Linux) yesterday.

I do have some hosts which are not showing corruption. I will see if they reset over the next 24 hours.

*EDIT* Detach/reattach worked for some hosts, have nice new host records which are correct. Can't merge them with the "old" host records though as the corruption affects the CPU and Operating System fields.
8) Message boards : Number crunching : Best way to remove old hosts (Message 2929)
Posted 1 Oct 2004 by TPR_Mojo
Post:
It would be nice for all BOINC projects (and hey yes I know this is priority~ 7bazillion+3 ) to "archive" known dead hosts, by the user, thus keeping the total (stored on the "user" table?) and the host records if required.



©2024 CERN