[UNRESOLVABLE] - memtest86: How long for 128GB RAM?

Issues related to applications and software problems
Post Reply
User avatar
warron.french
Posts: 515
Joined: 2014/03/27 20:21:58

[UNRESOLVABLE] - memtest86: How long for 128GB RAM?

Post by warron.french » 2020/09/10 13:01:07

I am running memtest86 from media on a server that we are having problems with in our Production environment.

Technically, we are running RHEL-7.8; but our problems have been around for the better part of a year from what my teammates tell me. I have only been on this job for 6 mnths.

Anyway, I am running memtest86, as suggested by our vendor to determine if we have hardware problems.

The specs of our servers, they are all provided by the vendor along with their specialized software, are:
  • 2 CPUs @ 8cores/cpu; with hyper-threading enabled
  • 16x 8GB (total 128GB RAM) at DDR4 @ 2133 MT/s


So. I have two really basic (I hope?) questions that I hope someone can assist me with:
  1. How long roughly should it take to run through all passes/tests/etc../etc... to complete the execution of RAM testing by memtest86?
  2. Does memtest86 run any other diagnostics? Or is it simply doing memory testing very well?
Last edited by warron.french on 2020/09/23 22:57:50, edited 1 time in total.
Thanks,
War

tunk
Posts: 768
Joined: 2017/02/22 15:08:17

Re: memtest86: How long for 128GB RAM?

Post by tunk » 2020/09/10 13:34:46

AFAIK memtest86 only tests the memory, but I think it uses a bit of CPU to do it.
You could also try to reseat all cables, DIMMs, cards, etc. a few times.

User avatar
warron.french
Posts: 515
Joined: 2014/03/27 20:21:58

Re: memtest86: How long for 128GB RAM?

Post by warron.french » 2020/09/10 14:46:24

Thanks @tunk.

That's what I thought, I just wasn't sure because the vendor was implying that memtest86 was a hardware tester. Which in essence is still true, but not in the same spirit that he was implying.

The tests/test is still running after 16 hours, so once it has completed on our server I will write back with an update.
Thanks,
War

User avatar
TrevorH
Forum Moderator
Posts: 29493
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: memtest86: How long for 128GB RAM?

Post by TrevorH » 2020/09/10 15:18:17

You'll want to leave it to run several passes and a pass can take hours.
CentOS 6 will die in November 2020 - migrate sooner rather than later!
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 is dead, do not use it.
Full time Geek, part time moderator. Use the FAQ Luke

User avatar
warron.french
Posts: 515
Joined: 2014/03/27 20:21:58

Re: memtest86: How long for 128GB RAM?

Post by warron.french » 2020/09/10 20:17:09

Yep, still running. It has been running for nearly 23 hours 21 minutes!

My boss told me to break it tomorrow morning and see if we can run a smaller set of tests that will not exhaustively test all of the RAM DIMMs, registers, addresses, and etcetera.

Do you know... can I run just Test #1, #2, and #3 only? Is there a means to implement testing with a selection of tests?
Thanks,
War

User avatar
TrevorH
Forum Moderator
Posts: 29493
Joined: 2009/09/24 10:40:56
Location: Brighton, UK

Re: memtest86: How long for 128GB RAM?

Post by TrevorH » 2020/09/10 22:30:56

There is and it tells you on the bottom of the screen what keys to press. But realistically it takes as long as it takes and you can't hurry it up without compromising the accuracy of the results so trying to it it in smaller chunks is really just going to take longer.
CentOS 6 will die in November 2020 - migrate sooner rather than later!
Info for USB installs on http://wiki.centos.org/HowTos/InstallFromUSBkey
CentOS 5 is dead, do not use it.
Full time Geek, part time moderator. Use the FAQ Luke

User avatar
jlehtone
Posts: 3044
Joined: 2007/12/11 08:17:33
Location: Finland

Re: memtest86: How long for 128GB RAM?

Post by jlehtone » 2020/09/11 10:00:45

TrevorH wrote:
2020/09/10 15:18:17
You'll want to leave it to run several passes and a pass can take hours.
Physics. If there is a clear, repeatable error, then it will show on every pass.

Alas, some errors are less obvious. They might occur only once in a blue moon, like nuclear fission. To find them one would test until the error is detected. That might take infinite passes. In practice you can't do that. Just run enough passes to feel statistically comfy.

You need to run at least one full pass to know how long it takes.


There is a community of "overclockers". They tweak the hardware beyond specs. Then they run some benchmark/stress test to see if the system remains stable under load; does not crash, does not boil, does pass tests (like calculating primes?).

EPEL has package 'stress', but it creates just a simple load.

BShT
Posts: 278
Joined: 2019/10/09 12:31:40

Re: memtest86: How long for 128GB RAM?

Post by BShT » 2020/09/11 11:22:47

try to test every single memory at once

its faster, if you are lucky you will find the problem before the last test

and using dual channels you will never know which memory is bad looking at addresses

User avatar
warron.french
Posts: 515
Joined: 2014/03/27 20:21:58

Re: memtest86: How long for 128GB RAM?

Post by warron.french » 2020/09/11 12:25:02

@TrevorH, I tried to run just 3 of the 13 tests, and I still see that Test #5 is being executed. That does not make sense at all.
Thanks TrevorH.

@jlehtone, I like your physics reference. My degree is in physics & mathematics. Not that I ever got to use them.
Thanks jlehtone.
Thanks,
War

Post Reply

Return to “CentOS 7 - Software Support”