Web Hosting Talk : Web Hosting Main Forums : Dedicated Server : What Causes RAM to Fail?

[RJohnson23]

I have a dedicated server that's been giving me a lot of trouble lately. Last week it began crashing and would kernel panic every time it would attempt to boot. The techs ran memtest86 and found that one of the RAM modules didn't pass and they replaced it. The server worked fine for a few days and now today I started getting the same problem and again, another module failed memtest86, which they replaced and now the server seems fine. Am I just incredibly unlucky or is there some deeper underlying issue?

[gordonrp]

Heat, power surges, poor quality RAM, other factors. It could also be a failing/faulty memory controller/motherboard.

[rds100]

Heat, bad power supply, cheap motherboards with blown capacitors, overclocking... all these can cause RAM to fail.
What motherboard is this? I guess you are not using ECC RAM?

[Mike_GoDaddy]

RJohnson23,

Have you had your provider test all the hardware or have they just tested the RAM? As gordonrp suggested you may have other items failing or a overheating issue. If you haven't already, have the data center check all the hardware on your server.

[funkywizard]

It's not unusual if one ram stick is bad that another is bad at the same time, so if things seem to be running solid from here on out that would be a good sign. Personally, we find it to be too much hassle and cause too much downtime to test a customer's ram while the customer is using their server. If we suspect bad ram, we would replace all of it, and then test whatever we took out. If the suspect ram passes our tests, maybe it's good, maybe it's not. If a specific ram stick fails our tests, great, we found out which one was bad and the customer got up and running without having to wait for us to find out. If the customer still has the same problem after we've replaced all of their ram, odds are that the ram we removed is working perfectly. The same kinds of problems that can make you suspect bad ram can also be caused by a bad motherboard, so at that point we'd also replace the motherboard, again, testing the suspected bad motherboard separately rather than in the customers working system. If we've replaced both the board and the ram and the server is still acting up, then it's unlikely to be a hardware problem at all. At least half the time, we can exclude failing hardware as the source of a problem, simply by replacing the suspect hardware in the first place. In many cases, no amount of testing can make you certain that a given piece of hardware is at fault, but if you replace it, and the problem goes away (or doesn't), that's the strongest evidence you'll get.

[srider]

Handling RAM without static protection can cause a failure that may not appear for a week or two. If your server has recently had the RAM upgraded that could be a factor.

[p0go]

Your best bet unless you have access to the server which I believe you don't, is to call or submit a ticket to the data center and report that you believe there is hard ware related problems going on and you want one of the techs to do some tests and checks and provide you back the results of it, and if most likely they will be able to tell you what the issue is, and get it fixed up for you.

If it is hardware related and your only renting a dedicated server, and you dont have access to the data center or managed-hands on-site at the data center there's nothing you really can do much more, besides have them check it on-site, physically checking the server is your best bet and to do some tests on the machine and see what happens.

Good luck and I hope your able to get it fixed. A buddy of mine had a issue something like yours, and it actually was bad hardware, and he was "ripped" off in a way, they told him it was a brand new server and such, and he asked for some managed-hands, and also had the server handled himself, and then found out that the server that was provided to him, was used by another customer that canceled a week before he ordered, and all they did was just power the machine back up, swap in a new hard drive, and installed the OS, and provided him the details. What actually gave off that clue was they didn't even give him all new IP addresses. They had to then replace the whole server with a brand new server due to some major problems and faulty hardware that was causing his problems.

He was lucky though, that they did at least put in a new HDD when they did provide him the first server or else he would of lost his data too that he admitted he didn't have a recent backup of it, and would of lost alot of work and important files and code, and the drive was revocable and was able to just pop out and pop it into the new replaced server, for him to then migrate the data off the old drive onto some brand spanking new hardware that ran fine.

He's still with them today after actually meeting with the company a month or so later and got to tour the facility and actually see his server which was cool, after he had the ticket escalated all the way up to the company President and CEO. He's now working with the company too, Funny how that worked out too.


Anyways, good luck and I hope your DC is able to get you back running good and normal.

[Krazy]

most of time cheap builds

[RRWH]

Cosmic radiation.

Seriously, this is actually what Cisco used to say all the time int the early 00's whenever we had RAM Issues - They would typically say SEU (single Event Upset) caused by Cosmic radiation.

Most common problem is actually poor handling of the RAM modules at some time in their life, which leads to eventual failure.

[Preet Arora]

Originally Posted by rds100

Heat, bad power supply, cheap motherboards with blown capacitors, overclocking... all these can cause RAM to fail.
What motherboard is this? I guess you are not using ECC RAM?

plzzz told me that i m using intel dh611ww,kingston ram and my ram is currupte today 1st time when i replaced my fist ram.......pc works so fine and today it happened agian my second ram is stop working.......plzzz help me........why it happend again and again

[SenseiSteve]

More often than not, memory failures are related to poor handling and should be replaced as a set, as mentioned above. Of course, power surges, spikes, AC ripple and heat can cause failures as well.