T610 getting random BSOD 0x000000f4 (0x00000003)

[Stuckeya]

Isdera,

A WES thin client is designed to have the Write filter running when it is in production use. The write filter should only be disabled for short periods while being updated by an Administrator.

Running in production without the Write Filter enabled is not supported.

You will find that temp files are pointed to the Z: drive as it a RAM Disk and therefore is very fast. Writing to a Flash drive with the write filter off is relatively slow. This can impact user experience depending how the device is used.

With the write filter on, writes to the C: drive are captured in the write filter which has a finite space. Too many writes will fill up the write filter. This may result is Blue Screen, but results are "undefined".

With the Write Filter on, the device is "guaranteed" to bring the device back to know state after power cycle/reboot. This is very powerful feature of Windows Embedded/Thin client implementations.

The current WES7 image  (V286) supplied by HP sets the RAM disk to 128 MB. I have seen needs for higher, but most implementations dont even come close to this.

Regards

Andrew S

[Powersurge · Accepted solution]

Being the originator of this thread, i thought i'd mention how i got around my issue to help anyone else that migth have this issue.

3-4months ago, we instructed our qty16  T610 thin Client users to NOT use the blue usb ports behind the Thin Clients.

After doing this,   BSOD's are gone!

Those  blue usb ports, you guessed it, USB 3.0 ports, seem to be creating the BSOD's on our end.  (even with latest driver dated July 2013)

I have a user today that had a BSOD.   Sure enough when i asked them to see if they were using the "blue ports" the answer was yes.  Someone had installed a wireless mouse on the USB3.0 blue ports over the weekend.  I removed it, put it on the regular USB2.0 ports (the black ports) and rebooted.   All is good now.  (and i'll report back if BSOD's show up again)

[Stuckeya]

Thanks heaps for the update.

I must say that I am not a big fan of USB3 on thin clients as nothing in a Client Virtualisation session needs the sort of speeds that a USB3 provides.

I have seen some devices not work real well either. One even caused the device to hang in BIOS when connected to USB3.

BTW, please check your BIOS is up to 1.16 - 3 Jul 2013

I would ask that you PLEASE log a case with HP support, so the information (WES7 version, BIOS version) can be captured and addressed. HP do take these issues seriously but need customer information to create a fix.

Regards

Andrew S.

[Powersurge]

I have a case with HP. case no: 4710196036 BSOD and yes, i do have the latest BIOS.

[Stuckeya]

And the WES7 verson is V290??

[Meilink]

Frankly, the write filter story is **bleep**. Those flashdrives are just mini PCI-e solid state drives. We even managed to swap one out for a bigger one for testing purposes a while ago.

We have 110 T520's with WES7 running for 2 years now. They all have their write filter turned off because we couldn't get single sign on working for Citrix, the login webpage needs a cookie to perform single sign on and that cookie gets altered after every login. There for, the write filter would prevent people from logging in.

My home pc has an 128GB Crucial M4 SSD in it, obviously no write filter on it, has been running flawlessly for almost 5 years now.

Yes, any flashdrive deteriorates over time (guess what, so do traditional harddisk drives), but I see some people on the internet say they will last for about 100.000 rewrites, that would mean such flashdrives last 2 months, which would just fall under warranty.

The unit in our T520's is a 16GB Sandisk U110. Sandisk itself claims a 2 million hour (just over 228 years) lifespan and/or up to 10TB of total data written to the unit. That means you can rewrite the entire disk 625 times before it will start to fail. U110 data sheet here

Could I maybe direct you to this article on techreport.com, where they put 5 SSD's through an endurance test to the death, the first SSD failed at just over 600TB of writing, the last one gave out at just over 2.4PB