Client errors with einstein_S5R5_1.01_i686-pc-linux-gnu
23 Apr 2009 16:40:04 UTC
Topic 194309
(moderation:
Hi.
I've recently restarted my work for E@H, but I have a big problem : all my computing ends in "Client error". This is what I saw in the logs :
dim 19 avr 2009 06:00:55 CEST|Einstein@Home|Computation for task h1_0632.45_S5R4__186_S5R5a_3 finished
dim 19 avr 2009 06:00:55 CEST|Einstein@Home|Output file h1_0632.45_S5R4__186_S5R5a_3_0 for task h1_0632.45_S5R4__186_S5R5a_3 absent
dim 19 avr 2009 06:00:55 CEST|Einstein@Home|Starting h1_0632.45_S5R4__174_S5R5a_1
dim 19 avr 2009 06:00:55 CEST|Einstein@Home|Starting task h1_0632.45_S5R4__174_S5R5a_1 using einstein_S5R5 version 101
dim 19 avr 2009 06:01:57 CEST|Einstein@Home|Sending scheduler request: To fetch work. Requesting 3048 seconds of work, reporting 1 completed tasks
dim 19 avr 2009 06:02:02 CEST|Einstein@Home|Scheduler request completed: got 0 new tasks
dim 19 avr 2009 06:02:02 CEST|Einstein@Home|Message from server: Project is temporarily shut down for maintenance
Can anyone help me with this ? E@H used to work fine, I also run S@H and PrimeGrid without any problem, so I must have done something wrong...
Thanks.
--
Nicolas FRANCOIS
We are the Micro$oft.
Resistance is futile.
You will be assimilated.
Client errors with einstein_S5R5_1.01_i686-pc-linux-gnu
)
Can it be clearer that this? There is nothing wrong, only the problems with the DB, and they are working very very hard to solve this. All can be read on the projects main page, and several treats.
Greetings from
TJ
RE: RE: dim 19 avr 2009
)
Well, no, if you look at the host page, you'll see that the host in question actually managed to report a few results yesterday, and yes they are all client errors. This is unrelated to the outage.
Ids the host in question perhaps overclocked?
Another possibility is that a certain bug in the Linux kernel is causing this. What kind of Linux distribution are you using?
CU
Bikeman
RE: Well, no, if you look
)
Hi Bikeman,
I looked there before I gave my comment, and could (can) only see compute errors. After thinking longer about the problem, this has indeed nothing to do with the DB. However I got one or two client errors myself in the period of the outage since March, and jumped to conclusions to fast, excuses for that, to Darth as well.
TJ
Greetings from
TJ
RE: RE: RE: dim 19 avr
)
No it isn't. It's a genuine "Intel(R) Pentium(R) Dual CPU E2200 @ 2.20GHz".
Well, this is kind of a personal distribution : Linux From Scratch (http://www.linuxfromscratch.org), with a 2.7 GlibC, a 2.6.24.4 kernel and Gcc 4.2.3. Everything is compiled with no specific optimization, so it's not a stupid -O17 option to the glibc which could set my desktop unstable...
I saw that the kernel version was sometimes an issue in clients errors, is mine OK ? I'm going to change anyway, going to a new LFS with a 2.6.28.8 kernel.
Should I stop my E@H work in the meantime ?
Thanks for your fast help.
\bye
PS : how can I set a signature on my account ? I didn't see the link to the page where I could modify this...
--
Nicolas FRANCOIS
We are the Micro$oft.
Resistance is futile.
You will be assimilated.
RE: RE: ...Is the host in
)
Could it be a temperature problem, even without overclocking (dust bunnies)? All tasks I checked had
That's a service of the various stats sites. If you register for instance with BOINCStats, you can find it on this page.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
RE: Well, this is kind of
)
Ahh, ok, that looks like a smoking gun. The crucial point is the Kernel configuration setting CONFIG_PREEMPT=y which can trigger the error for certain kernels.
The problem was fixed after a Einstein@Home user submitted a bug report and helped track the bug down, the fix was first submitted for Linux 2.6.26-rc5 so your kernel should be affected.
I'd advise to update the kernel before continuing E@H.
CU
Bikeman
RE: The problem was fixed
)
OK, then I'll stop E@H, install a new kernel (it will anyway be usefull for my next LFS system :-), and restart E@H after that.
Thanks Bikeman ! Climbed some mountain lately ?
\bye
--
Nicolas FRANCOIS
We are the Micro$oft.
Resistance is futile.
You will be assimilated.
RE: I'd advise to update
)
That did the trick. Thanks Bikeman. It works now.
\bye
--
Nicolas FRANCOIS
We are the Micro$oft.
Resistance is futile.
You will be assimilated.
Good to hear that, Happy
)
Good to hear that,
Happy crunching
Bikeman