Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Comment on rpitc v2- 1.4 and locking problems
18-11-2015, 07:40 AM, (This post was last modified: 18-11-2015, 03:55 PM by greavette.)
#1
Comment on rpitc v2- 1.4 and locking problems
Hi All,

I've just installed rpitc v2-1.4 in our office workstations to replace a version I built for the Pi B back in 2013. I've been testing this version for about a month before deciding to install it at our workstations but over the past few days I've had more than a few lockup issues with our thin clients that has me concerned. My testing showed no locking issues which is why I took the plunge and installed it on our production workstations.

Out of 14 Thin Clients now in place I've had 4 lock up in the past 3 days. Once they locked up the first time I removed the overclocking on all 14 thin clients thinking that might be the problem but 2 have since locked up after overclocking was removed.

Our version we used on the Pi B from 2013 was rock solid. It was slower but our users had no issues. I'm not sure if this issue we are facing is due to rpitc or due to the Pi2 itself not being as solid as product as it's predecessor? I'm hoping this community can give their opinion on what the problem might be and possible solutions.

I use 2amp power supplies for all 14 clients. Some of my clients do have a daughterboard mounted ontop for a powerbutton. I usually have a mouse, keyboard and barcode scanner attached to each Pi workstation. I don't think the peripherals are taking too much power.

I'm now considering a few options that I thought I would share with this community for comment:

- A cron script to run daily (at night) to force a reboot of the Pi. I have No evidence that this will stop a P2 from locking but perhaps it will 'keep it fresh' so it won't lock?

-Using the onboard watchdog timer. I've tried this in the past and had issues with it getting stuck in a reboot loop. Perhaps if there was a way to set the timer so it takes longer before a reboot is done to help it not getting stuck in a loop?

- An external watchdog timer. I don't want to have to buy another external board for a cheap thin client like the Pi...but if it keeps our workstations healthy and not locking I'll consider it.

Does anyone else have comments/thoughts on these locking issues and what might be the best solution?

Thank you.
Reply
18-11-2015, 11:23 AM,
#2
RE: Comment on rpitc v2- 1.4 and locking problems
After the freeze the i/o green led is always on?
Reply
18-11-2015, 02:15 PM,
#3
RE: Comment on rpitc v2- 1.4 and locking problems
Yes I believe so...but I will confirm with my coworkers the next time it happens.
Reply
18-11-2015, 03:46 PM,
#4
RE: Comment on rpitc v2- 1.4 and locking problems
After the freeze did you need to rewrite the image?
Reply
18-11-2015, 03:51 PM,
#5
RE: Comment on rpitc v2- 1.4 and locking problems
Sorry for the OT.

You mention: "....Some of my clients do have a daughterboard mounted ontop for a powerbutton...". Since I have the same idea, can you tell me which one you use and your experience?
Reply
18-11-2015, 03:55 PM,
#6
RE: Comment on rpitc v2- 1.4 and locking problems
Hi Gibbio,

No, I've only hard rebooted the Pi and it starts up again. If the same workstations have the problem again I'll either re-image or use another card.

Thanks.
Reply
18-11-2015, 06:43 PM,
#7
RE: Comment on rpitc v2- 1.4 and locking problems
Hi Walter,

I use a Remote Pi Board Plus 2015 from MSL Digital Solutions - http://www.msldigital.com/

I've been testing their boards for the power button since March 2015. Matt from MSL Digital is very helpful and responsive to suggestions and has implemented some features to the firmware we needed (long press to shutdown the Pi if it locks up and secondly, upon power failure the firmware now allows the Pi to autostart when power resumes). Both these features were critical (in my opinion) to be useful in our office. Matt was kind enough to add these features in the fall of 2015 so I purchased 10 and have been very happy with my testing. I've now installed 8 of these in our office and it's a nice addition to have for turning off/on the Pi if the need should arise. I do instruct the employees that they should never turn off the Pi's unless it is frozen. But users have other ideas sometimes no matter how much I educate them. In the past I had provided power to my Pi Thin Clients using a powered USB Hub with an on/off switch. This wasn't good in my opinion as it just "cut" the power which could corrupt the SD card.

One word of advice for you if you do look into these daughterboards for your Pi thin clients. In my testing the case MSL Digital sells is not good in our business environment. The tabs that lock it together are brittle and can easily break which happened to me (and I'm being careful). I can just see my co-workers getting frustrated trying to open it up and breaking the case. So I did more research and testing and settled on a case from Adafruit which I then modified. I used a stepped drill bit to open a hole at the top of the case for the power button. I then added a hole at the side of the case for the new usb power (it's right above the current USB power on the Pi 2). And I opened up a small hole at the bottom corner of the case where under the bolt head which is used at the corner of the Pi to keep the daughterboard up. You don't have to go through what I did if you like the case from MSL digital..but if you are interested in what I did to modify the case from adafruit I can send you pictures if you like.

Let me know if you have any other questions regarding the board..I'd be happy to share.
Reply
19-11-2015, 07:16 PM,
#8
RE: Comment on rpitc v2- 1.4 and locking problems
Hi Gibbio,

One of my 14 thin clients is still experiencing freezing/locks whereby the only fix is to hard reboot the pi thin client by pulling the power. I've confirmed that this thin client (like all the others) has overclocking turned off. Next steps is to put in a new SD card and see if this fixes the freezing problem. If it continues to freeze I will replace the Pi2 hardware itself with a new one.

If you have any ideas on what more I should be looking at with regards to possible freezing/locking issues please let me know. I'll keep you posted through this thread on further updates.

Thank you.
Reply
19-11-2015, 08:42 PM, (This post was last modified: 19-11-2015, 08:45 PM by greavette.)
#9
RE: Comment on rpitc v2- 1.4 and locking problems
Ok everyone...I'm doing as much research as I can on this issue. I've got a lead that I would like to bring to this community's attention for review and comment.

I was reading in this post about others having freezing/locking issues that seems to be related to USB - https://discourse.osmc.tv/t/solved-usb-h...smc/6230/6

The gist of their conversation is around the hot swapping of USB devices from a Pi2:

Quote:The fact is that the vast majority of USB "chargers" are not well regulated, and drop well below 5v before reaching their rated current. This is proven over and over again when put to the test. The Pi (and Pi 2 in particular) is much more fussy about the voltage than a phone or tablet is when charging. Remember that a phone or tablet has a battery to help run it (which effectively acts as a really large filter capacitor) while the Pi relies entirely on the external supply and still has undersized filter capacitors.

The post then goes to explain a bit more about the work done on Pi B+ and Pi 2 with regards to USB devices:

Quote:In the Pi B+ and Pi 2 B the power supply circuitry was completely reworked and as well as now having a switching regulator instead of linear the USB ports are now isolated via a current limiter which can be switched to 600mA or 1.2A in software (max_usb_current in config.txt)

In the default 600mA setting this means that if a USB device draws too high a surge of current when initially connected it will typically not crash the CPU, but it often will crash the USB device itself. If you enable max_usb_current=1 the limit is increased to 1.2A (combined for all 4 USB ports) which can help the USB device itself not to crash when connected but if the power supply is not sufficient it will increase the chance that the CPU and thus the whole system crashes.

So on a B+ / Pi 2 hot swapping of USB is much more reliable but it can still cause hangs no matter how good your power supply is.

This got me to thinking. The USB devices on my one thin client that is freezing...perhaps the freezing is not the O/S, but a disconnect of one or more USB devices which results in a crash of the USB device. To us it looks like a freeze since the mouse and keyboard doesn't work anymore. At the time the apparent 'freeze' happens I'm not onsite and not checking if ssh access is still available. I think what I'll do next is add this particular thin client to my OMD monitoring system and see if reports from the users of "crashing" is corresponding to a 'Offline' status in my monitoring system.

I see in the rpitc config.txt file that there is not setting for max_usb_current. Has anyone tested (Gibbio perhaps) what setting this value to 1 will do on the rpitc thin client?

I'm also not using a USB Hub for my USB devices (the three USB devices..mouse, keyboard and hand scanner) are all plugged directly into the Pi 2 thin client). If locking still occurs I will move these devices to a powered USB hub instead.

Any and all comments are welcome.
Reply
19-11-2015, 09:43 PM, (This post was last modified: 19-11-2015, 09:51 PM by Walter Stemberger.)
#10
RE: Comment on rpitc v2- 1.4 and locking problems
(18-11-2015, 06:43 PM)greavette Wrote: ....

Thank you for the info. I am using a TEKO case in this moment, I have to see if it is compatible, and with which modifications, to this add-on.

I had some issues with the USB systems, but I did not studied them very well, since I connected them to overclocking. The system was not freezing, but it lost the IP address. Looking at the log, I found that the USB subsystem was restarted, but the system never got (or even asked, I do not remember) the IP address.

Since after commenting the overclocking lines the problem apparently disappeared, I never entered in it deeply.
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)