Server crashing

Celegorm

Site Team
Staff member
Messages
11,741
Location
USA
So looking for opinions here.

I have a server on a rack in my basement that I use for plex. Recently it's started crashing. Nothing shows up in the event viewer to indicate a problem, no BSOD shows on the monitor I have hooked up to it and when it crashes the numlock/capslock key do not change the indicator lights on the keyboard. The fans and all other indicator lights still are active/on.

Now I had seen this before when a PSU was crapping out it'd keep the fans on but everything else died so I swapped out the PSU Sunday with a known-good unit, reseated the RAM and ran memtest (which passed). This hasn't helped.

The crashes are random. sometimes within 30 minutes of booting, other times it can take 2 days to crash again.

My plan now is to start taking out ram sticks as some of them are new to see if that helps but I'm looking for other opinions/experiences as I start to test.

Full specs:
Asrock ATX DDR4 X99 Extreme4
6 x 8GB sticks DDR4 2133
intel Xeon CPU - 18-core at 1.8 Ghz
HighPoint RocketRaid SAS Card
8 Hard drives.
EVGA 8400 GFX card
 
Last edited:
How long did you run memtest? If you can afford to have your server down for a while, I'd run it continuously for a good 8 hours.
 
It was about 6 hours or so when I ran it - made it through 2 passes before I rebooted. The hard part is finding time to have it down. During the day i'm streaming plex to my office for music, over night my PCs backup to it's disks. I could probably verify backups and re-run Memtest overnight but I did plan on removing 2 sticks (they're all sold in pairs so 1 bad one means I'm sending 2 back) tonight to start with that. I could postpone and re-run memtest for you.
 
Naw, if you're going to swap the memories anyway, then go ahead as planned. It really is most likely you have a bad module, considering your symptoms.
 
If it was a RAM/CPU issue I'd have thought error codes would show up in the software.

My bet would be on a power related issue, tried hooking it up to a UPS and seeing if it goes off then? Possibly a dry solder joint of some description on the motherboard? Something shorting out? Any plugged in peripherals being removed could help diagnose the issues a bit further, sounds a bit of a strange one.
 
Is the OS crashing or simply the Plex server?

I run Plex on my NAS and it will occasionally stop responding (clients timeout when they attempt to connect). I simply shut down the Plex server and restart it and everything works again. It seems to happen mostly when I have been adding new content to the libraries.

Are you running the latest release of the Plex server? There was a new version released about 2 weeks ago.
 
It's hooked up to a UPS and the only 2 peripherals that are plugged in are a keyboard and monitor.

To answer your question strollin, it's the full OS not just Plex. I'm unable to remote in or access any web-based servers (it's server 2012 Essentials so it has a built-in website).


And to expand upon what I said with the event log - it's just like it dies. It'll be logging things as normal then just stop. No errors or warnings to indicate an issue just a halt in all logged activity.
 
And to expand upon what I said with the event log - it's just like it dies. It'll be logging things as normal then just stop. No errors or warnings to indicate an issue just a halt in all logged activity.

This is the part that suggest to me that are right in suspecting the RAM. It's possible that the CPU is failing, but in all my thirty years of messing around with computers, I've only seen a CPU fail twice.
 
This is the part that suggest to me that are right in suspecting the RAM. It's possible that the CPU is failing, but in all my thirty years of messing around with computers, I've only seen a CPU fail twice.

And I've only seen it fail once. It's a new CPU and never over-clocked so the odds are very, very low. I'd put it bad RAM or motherboard before bad CPU.
 
Back
Top Bottom