Solved BOSD every week on a server up 24h, maybe hardware degradation?

[GiovanniG]

Hi mates, thank you really a lot for your kind opinion.

I've a server for videosurveillance record and cameras mosaic display, it's up about 2 years, here's the story:
after some months it "freezed" randomly (about once a month) (only mouse was working) when processing every night a scheduled file deletion on some folders of snapshots (it deletes the files older then 10 days): I noticed it happened when there are a considerable amount of file, around 50.000 (camera with wrong settings, too sensitive, too many snapshots).

The server was working about at 70% of processor, the temperature was about 73°. Some crashes with BOSD may happened as well after 1,5 years, but too rare to be worried about.
After a main crash of motheboard (mosfet neat CPU burned, cause cooler was not cooling the motherboard), with Mobo replaced, cooler replaced (right type), and mosaic set up to lower cpu consumption (now it's about 12%) I have 70°.
The server is up 24h, scheduled reboot after 3-4 days a week to try to reduce BOSD (if may happen anyhow), drivers are up to date (Nvidia, Ethernet.. the most used), Mobo is Asusrock 970 Pro3 2.0, cpu AMD A8, powersupply is from a good brand (not cheap one), Windows is almost up to date, no overclock, Video settings are at lowest value (triple buffering, etc.)

I try to understand what is wrong with it, maybe power supply? 3Vsb:3,42V, 3Vcc:3,31V, VIN1: 2,04V, VIN2: 1,73V, VIN3: 2,04V. Vcore: 1,42V, 12V: 11,99V, AVcc: 3,33V.
Maybe video card with elettrolitic condensers degraded?

Thank you for helping me find the cause

 

[GiovanniG]

It may be possible that service center didn't place the thermal paste in the proper way on cpu? ther may some areas which are not in good contact with cooler.
In your experience, what can make crash a Win7 64 after 4-7 days?
There is another important info I would add, let's say that only 2 times on 10 the system reboot itself and Windows runned as before, usually I advice the problem because the server stops answer ping and I got an email from the twin server, when I switch on the monitor I see absence of signal on HDMI, it looks like the reboot occurs but the BIOS froze somehow.. it is necssary to press the reset button or switch off the power (long pressing on power button). This may indicate something.. on hardware..
The BIOS is updated to the latest, all settings in BIOS are default.

 

[samuria]

see if this tells us anything

Please download MINITOOLBOX and run it.
Downloading MiniToolBox

Checkmark following boxes:



Flush DNS
Reset FF proxy Settings
Reset Ie Proxy Settings
Report IE Proxy Settings
Report FF Proxy Settings
List content of Hosts
List IP configuration
List Winsock Entries
List last 10 Event Viewer log
List Installed Programs
List Users, Partitions and Memory size
List Devices (problems only)



Click Go and post the result.

 

[axe0]

There is another important info I would add, let's say that only 2 times on 10 the system reboot itself and Windows runned as before, usually I advice the problem because the server stops answer ping and I got an email from the twin server, when I switch on the monitor I see absence of signal on HDMI, it looks like the reboot occurs but the BIOS froze somehow.. it is necssary to press the reset button or switch off the power (long pressing on power button).

Is the BIOS always frozen when this happens?
That may indicate to a possible defect on the motherboard.

 

[torchwood]

note
Windows updates has failed snce 9/1, (earliest date in log), this is due to the fact that your time zone is failing to sync.
Your system states Italy - Time zone set to Russia.

Roy

 

[GiovanniG]

Thank you all for your kind answers!!

The system is not updating because I disabled the windows update service, I do updates after some months manually cause they may create some issues (higher cpu usage, fragmentation, dozen of thousands of files more on C:, etc.) by the way the problem still from about one year nevertheless I updated it, so this won't solve problems.

No, the BIOS not always froze after BOSD reboot. This is why I'm thinking the problem could be connected more with graphic adapter. The graphic adapter (which is "loaded" displaying 24h the mosaic) may stop working properly, this may cause a driver problem which drive Win7 to BOSD. The problem then still (because probably the BIOS doesn't reset the GPU) on BIOS first check.. and the BIOS stops like no graphic adapters detected/malfunction. This is just my suppose. It can be instead the RAM, the power supply who may produce an instant "lack of power", or any strong noise (like transients) at power source that are not prperly filtered by power supply.
By the way the twin server is connected to the same source, but it uses another power supply, which appear cheaper than this (who sold me the server "fake" me with that, but I discovered it only too late when for the first time I opened the server).

samuria I attach here the file.

From the minidumps did you discovered something? I suppose that all of them are caused from the same problem.. those may contain some useful indications? I've not unfortunately the skills to understand them

Thank you a lot!

 

[GiovanniG]

I would spend some hours testing RAM with memtest86, and some others stressing the GPU (benchmark). A failure may be indicative..

 

[axe0]

The dumps indeed point to the same, the Nvidia display driver.

Please uninstall everything of Nvidia using Display Driver Uninstaller and install new drivers from Nvidia. Be sure the clean install box is checked and only install the Graphics driver and the PhysX driver, you can use this tutorial to do so :ar: NVIDIA Drivers - Avoid Problems

 

[GiovanniG]

Thank you, it worth a try!, I got an idea.. I can switch the two graphics between servers, and see what going to happen with the other

 

[GiovanniG]

Good morning all!
I switched graphic cards on saturday at 19:00 the first automatic reboot occours at monday at 6:00 (after 35 hours) and today at 11AM (after 29 hours) I had the crash. It is not the graphic card, then it should b something else, RAM or motherboard. Next step I would like to test ram for some time and spread again the thermic paste.

There is a stupid setting in bios that shows me quickly a brief "INIT" of the hard drives, it doesn't show it on the other server. I guess this is because the system doesn't restart, it forzen somehow there. But of course I've to solve the stability problem.

You guys understood from dumps what is the reason? Have them all something in common?
Thank you!

 

[axe0]

All of them have something in common, a common issue with BSODs, memory access violations (for whatever reason caused it).

When you switched the GPUs, did you seat them in the same slot or different slot?

 

[GiovanniG]

the same slot, the only pci express available. The mobos and CPU are equal, even RAM modules.
What can't cause memory violation?
Power/defective power supply?
Wrong contact of the CPU cooler?

There is also a BIOS malfunction that occours after the BOSD.. it freeze and doesn't boot again. It may be connected to the SATA system, *may*. The thing is when it occours I'm not there to see the monitor, and when I switch it on the HDMI is with "no-signal", maybe after minutes the BIOS could itself put the GPU is power safe mode? Then I can't see where it hangs.

So briefly there are two problems:
1) unstability
2) when unstability occours, BIOS isn't proceeding with boot after.

Thank you

 

[axe0]

That the BIOS freezes means that either the battery needs replacement or the motherboard is malfunctioning. At least, those are a few basic options why the BIOS could freeze.

 

[GiovanniG]

Hi all!
I had some time to run Memtes86 and repeat it.. in 30 % of cases I noticed a RAM fault at the same address.. about position 2,81GB.. and only there! Then I copletely switched the RAM modules between servers to see if it was a RAM fault and it compared again. So, like someone said here, the problem is on motherboard! I still can't believe how motherboard can fail a specific ram location only.. for me it's a mistery!
Fortunately the board is under guarantee and I'll ask for a replace.

Also, I disabled the AMD IDE settings under BIOS, which created that "INIT" page on display during BIOS boot, and tonight the BOSD "a" appeared again, but the BIOS let it reboot.

 

[GiovanniG]

There is not big sense in your words, sorry, a pin can't be assigned to that small area of ram only, if it is broken the problem will recursively reflect to the other areas of ram too. Plus, as I wrote, the problem doesn't compare at each memory scan, but appears only some times, less about 3 times every 10 test restart/system reboot, and with only some tests of Mem86, this is really confusing. With some tests I tested ram in that area for 40 minutes without revealing problems, we may suppose that bilions of data where written and read back in that "faulty" locations without issues, another time after a reboot as soon I started memtest86 I saw a scroll of 19 errors in a while..
Nothing broken here, it's a strange behaviour of northbridge, maybe connected with access times. In my opinion there is a parassite capacitance somewhere on the tracks of the board