w/u fail at end of processing
26 Aug 2007 17:21:10 UTC
Topic 193080
(moderation:
Hi, I have a linux machine running for ages processing w/u just fine.
I have now had 5 or 6 w/u end with
(process exited with code 193 (0xc1)) right at the end of processing.
The app ver is einst_s5r2 4.35.
Is this a machine prob?. Cannot find what code 193 is.
Ta
Nairb
w/u fail at end of processing
)
The BOINC Error Codes page (from the BOINC Wiki) says 193 is an ERR_INVALID_EVENT. Unfortunately, I don't know what that might mean relative to Linux and can't help you any further.
RE: Hi, I have a linux
)
The science client terminates with segmentation violations, in different stages of the computation. I would not rule out a problem with the hardware. If it's overclocked, you might consider running at design speed for some time to see if that fixes the problem.
CU
H-BE
Thanks for the replies. I
)
Thanks for the replies. I did check on the boinc wiki.... but plainly not in the right place. Each w/u fails at the end of processing. Which is frustrating on a 44hr w/u. I guess the science is lost as well as the credits.
So I have suspended ein for a while and downloaded some seti w/u to see if the same thing happens.
Maybe its time to run a mem checker for several hrs. This machine has been totally solid for years. This running flat out 24*7 sorts the good kit from the duff.
Nairb
RE: Thanks for the replies.
)
Not all fail at the end, some fail at the very beginning. Hmmmm....could be related to zipping/unzipping files ?? And it started with the new app version ....
You know, there's one other thing you should try: install the latest beta test version from the beta test page. Something related to zipping/unzipping was fixed in this beta release, it could well apply to your problem!
Good luck!
H-BE
RE: Not all fail at the
)
Is it by any chance overclocked? It could easily be that for your particular machine, the new ap version by bad luck represented a slightly more challenging proposition--so if you had adjusted to very little margin before, it could have gone negative.
For my Core 2 machines, the Einstein ap then current when I was adjusting them (in April and in early August) definitely was more demanding from a speed/voltage point of view than was the SETI ap or just staying alive in Windows, so this can happen.
I mention it mainly because your symptoms are a pretty good fit. When I was two voltage increments too low for Einstein success I got a failure about once a day. When I was one increment too low I got one failure in two weeks. In weeks of running, I had no SETI or system failures, even a couple of increments lower voltage than any of those.
Your system's error-reports are utterly different than mine, but you are running a different architecture CPU on a different OS, so that is hardly surprising.
RE: The BOINC Error Codes
)
Stick, positive error numbers are a problem with the application.
Negative error numbers are a problem in BOINC.
Error -193 is a BOINC error. This error number is no longer used in the present code.
Error 193 is a SIGSEGV (segmentation violation) error in the application, pointing to memory problems or the Einstein application making invalid memory references.
nairb, besides checking the memory, also check if you still got enough paging file left.
RE: Stick, positive error
)
Jord,
Thanks for correcting my screw-up!
Stick
Your comp was running OK with
)
Your comp was running OK with 4.21, then with 4.35 all your WUs are failing, then i have one more reason to dislike 4.35, LoL (not performing as well as 4.21 on Core 2)
Archae coud be right about the vcore thing, i had some Thoroughbred Athlons fail at stock volts/clock as they got older, they might require a voltage bump even without overclocking. Yours is a Barton core anyway (i think), they were considered to be better. Got a Barton 2500+ still going strong at stock volts, but it was never used for crunching, maybe thats how it survived this long? :)
Team Philippines
Well I think Archae is on the
)
Well I think Archae is on the right track here. This machine is the only overclocked pc I have. Its a barton core amd overclocked from 2500 to a 2800 and has been for the last 3 or 4 years. Its always been very stable with a stick of corsair xms memory. I never changed the core voltage as it did seti w/u fine. Its now done 4 seti w/u without fault. Its only with the new ein app that its started failing. Its not using any swap space either, so me thinks its time to retire this pc from ein work. I have put 2 older pc's on the job to see if they are ok with the new app.
Nairb
So I decided to try 2 older
)
So I decided to try 2 older pc's...... 1 of them is a dual 700mhz xeon machine.
Never been overclocked and has run without any problem on any project for at least 5 years. Except that is for the latest downloaded ein app.
The first 2 wu quit after 44 hrs with the good old 193 error. The last 2 quit at the end of processing 106 hrs!. The final wu I aborted!. This poor old machine is still doing rosetta/seti etc without fault.
The other machine is running win98. And has done 2 or 3 ein wu fine.
So the conclusion is... ein dosent like any of my linux machines any more.
So when it becomes less fussy I will join in again.
Nairb