Untangle crashes with some regularity

[yumbrad]

I am subscribed to Untangle Home. Love the features. But I have to power cycle the box every day or two when Untangle crashes. Sometimes it still passes traffic, usually it doesn't. But the webserver does respond, in various ways. I will attach pictures of two of those ways, if it lets me. Is this a known issue? I suspected RAM issues, but the way it crashes is almost always the same and I would expect more randomness with bad RAM. But just in case, I went down to one stick from four, and tried each stick for a few days - the same # of crashes no matter which stick. Is this a known issue?

Build: 15.1.1.20200905T150556.8ca624f09e-1buster
Kernel: 4.19.0-8-untangle-amd64
History: none
Reboots: 62 (58)
Current active device count: 23
Highest active device count since reboot: 48

Screenshot_20200823-103453_Chrome.jpg

[TirsoJRP]

Do a burning test on your system.

[sky-knight]

You either have a hardware problem, or hardware incompatibility.

Untangle doesn't "crash". When it does in the vast majority of cases it means you have a defective hard disk, or RAM.

[yumbrad]

You either have a hardware problem, or hardware incompatibility.

Untangle doesn't "crash". When it does in the vast majority of cases it means you have a defective hard disk, or RAM.

Ok, I'll try a burn-in test ... this is all new brand-name hardware, is there a list of incompatible hardware?

[sky-knight]

Ok, I'll try a burn-in test ... this is all new brand-name hardware, is there a list of incompatible hardware?

It's Debian 10 based, and Debian doesn't work like that. There is no hardware compatibility list. There are appliances with support, otherwise if you roll your own... well... it's your hardware you figure it out. That's how that works.

[Kryptonit3]

Enable SSH and the next time it crashes see if you can still SSH into the box. Maybe check out "top" or similar to see if any specific processes are overloading CPU/RAM, assuming also you have free disk space and the system isn't coming to a crawl due to low space.

[LaurentR]

Ok, I'll try a burn-in test ... this is all new brand-name hardware, is there a list of incompatible hardware?

I second that. I had Untangle crashes and it was marginal DRAMs. You should run DRAM testing for a few hours at temperature. What hardware do you have?

[yumbrad]

I second that. I had Untangle crashes and it was marginal DRAMs. You should run DRAM testing for a few hours at temperature. What hardware do you have?

This is all the hardware:

Gigabyte B450M DS3H (AMD Ryzen AM4/M.2/HMDI/DVI/USB 3.1/DDR4/Micro ATX/Motherboard)
AMD Ryzen 5 1600 6-Core, 12-Thread Unlocked 65W Desktop Processor
v-Color Prism RGB DDR4 3200MHz (PC4-25600) CL16 1.35V Desktop Memory (TL48G32S8KGRGB16)
XPG S40G 512GB RGB 3D NAND PCIe Gen3x4 NVMe 1.3 M.2 2280 Internal SSD (AS40G-512GT-C)
Intel EXPI9404PTLBLK Pro/1000 PT Quad Port LP Gigabit PCI-E Network Adapter
NVIDIA Quadro NVS 300 PCIe 2.0 x16 512MB HP 625629-001 632486-001
GIGABYTE PW400 80 Plus, Quiet Fan

[yumbrad]

Enable SSH and the next time it crashes see if you can still SSH into the box. Maybe check out "top" or similar to see if any specific processes are overloading CPU/RAM, assuming also you have free disk space and the system isn't coming to a crawl due to low space.

Ok, interesting. Got a crash this morning while on the computer and traffic stopped flowing. Went to SSH and saw that it was still logged in from last night, and still running top - chugging away, nothing looking much different - java usually taking the top spot, but similar CPU/MEM to last night.

Tried to access the untangle server via http and opened a couple browser windows.. Just sat there waiting for the server.

Decided to quit top and see what else I might see, and as soon as I quit top, both browsers got through! They showed this:
service.jpg

Went to run top again, and got this:
top.jpg

Any clues here, or is the randomness of bad RAM still the most likely culprit. I haven't had a chance yet to burn in, since we've been using the internet so much, and it only takes 20 seconds to reboot Untangle.

[jcoffin]

Another option is to pull one simm out at a time and run it for a while.