bad wu - Project: 6882, run 808 clone 0 gen 13

[betchern0t]

Hi,
I have a bad wu:

project 6882, run 808 clone 0 gen 13

It completes 145000 steps of 250000 runs the next chunk for a while then the terminal it is running in then disappears. Log states: corestatus 8B (139) Client core communications error 0x8b this is a sign of more serious problems shutting down. I caught a glimpse as the terminal window was disappearing of a segmentation fault error.

Box is brand new intel i5-2500 running ubuntu 11.04 in a KVM vm. The folding client is running in the VM. Please advise or point me at docco on what to do next.

Cheers Paul

[bruce]

Welcome to foldingforum.org, betchern0t.

It's unlikely that you have a bad WU, but we can't be certain yet. Error 0x8b is not a FAH error but rather an unstable system detected by the OS. It's generally due to overclocking/overheating or a memory failure. FAH only shuts down due to "more serious problems" when the errors come from the OS detecting hardware that is malfunctioning.

If you're overclocking, what benchmark did you use to insure stability? Have you thoroughly tested RAM? Is the VM you're running somehow interfering with the processing of the WU?

The stats server has no reports associated with this WU yet, but that's not unexpected. Your client is not uploading an error report and project 6882 has a Preferred Deadline of 14 days. Once your assignment times out, it will be assigned to someone else for completion. I'll mark this topic for followup.

[betchern0t]

Hi Bruce,
thanks for the response. I am not overclocking. The temps are in the normal range - 50 -60 degrees C with a rated max of 72.5. CPU core is generally 100% while the job is running. To check that contention wasn't a problem removed the other load - a series of boinc projects with the same result. In any cases I have assigned two of my four cores to the vm.

I am using the VM to supply a fixed resource for these kind of things rather than using the idle approach

I have completed about 5 or 6 jobs for folding at home so far. One thing I have just thought about is to up the RAM for the VM - currently sitting on 512mb. I will try that and report back. I haven't specifically run memory tests on the RAM - there is 8Gb - but may schedule for overnight.

Is there any way to cancel this WU so that I can continue? Early days and I am keen to do something useful.

Cheers Paul

[Grandpa_01]

If it is continually failing at the same spot it is possible it is a bad WU it is also possible it is just a tough spot in the WU that needs more resources than you are alloeing. But bruce is right error 0x8b is usually an hardware error I have found that it is usually associated with memory. 512 mb is not enough for a VM and folding I would give the VM 2GB of memory if you have 8GB available.

[betchern0t]

Hi Grandpa,
looks like you nailed it. It has now completed that section of the WU. VM has all four cores and 3Gb RAM.

Cheers Paul

[betchern0t]

crashed again on 155000....

[Grandpa_01]

If it made it past the previous failing stage it is most likely not a bad WU.
VM can be fitful at times. Are you using v6 or v7 of FAH

[betchern0t]

Hi Grandpa,
I have done 14 passes of memtest86+ with no errors. Memory testing is never definitive when no errors show but....

I am using 6.34. Being a newbie I went with what was offered.

Cheers Paul

[PantherX]

The WU Project: 6882 (Run 808, Clone 0, Gen 13), isn't a bad one as it was successfully completed by another donor:

Your WU (P6882 R808 C0 G13) was added to the stats database on 2011-07-23 05:05:15 for 69 points of credit.

[SombraGuerrero]

I've seen this in other threads as well. 6.34 was working perfectly for me until (big shock) the most recent round of glibc updates. I have installed, configured, and run nscd to no avail. I have reverted back to client version 6.02 which is what i did the last time this happened with 6.29 and I am once again successfully folding all cores.