Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
New Firmware Uploads
#1
I have added some more firmware uploads at http://files.star-os.com

Of particular interest is the beta section where I have War1B and Ventana trials using an older Driver with the modern additions.  It actually seems more stable than the newer drivers.

The only explanation I can think of is that the newer drivers have the failed attempts to fix the stall issue.  In general it is a good idea to remove code that does not fix and issue, but it seems this was never done, with the result of possibly new issues being added to compound the situation.

Anyway, I would be interested in the results.
Reply
#2
(07-13-2018, 12:23 AM)lonnie Wrote: I have added some more firmware uploads at http://files.star-os.com

Of particular interest is the beta section where I have War1B and Ventana trials using an older Driver with the modern additions.  It actually seems more stable than the newer drivers.

The only explanation I can think of is that the newer drivers have the failed attempts to fix the stall issue.  In general it is a good idea to remove code that does not fix and issue, but it seems this was never done, with the result of possibly new issues being added to compound the situation.

Anyway, I would be interested in the results.

WAR1B @ 2.4Ghz AP, with 4.8.2.0-9313 - the AP fubars.  Stops communicating, disconnects all clients.  A simple 'activate' solves it, and with the card watchdog active, the AP resets/activates itself.  There is no longer a counter, so no easy way to tell how often this happens.

It happens (as before) more often on AP's with more associated cliens, it happens more on higher traffic, and it happens more where we have more noisy situations.  So for example, where we have WAR1B's working as Point-to-Point units (1 associated station, fairly isolated from noise because of the directional antennas) we have PTP links with uptimes of 100+ days.  AP's with 10 clients might only reset 3 times per day, an AP with 15 clients might reset 10-20 times per day.  One of our APs with 24 clients fubar'd and reset 77 times yesterday.
Reply
#3
ninedd, I know you do not like the multi radio units, BUT, after a lot of code review it dawned on me what could be causing the situation.

During critical data structure access, ALL interrupts are turned off, to guarantee that only the current lock user will have access. Normally that user could be changed because of the process scheduler. That would explain why the situation is more prevalent with more users, especially if they are heavy users.

Would you at least try a Ventana with a new 9239 firmware? For the moment I have them on my store at http://store.gigabit-gecko.com for $225 and free shipping.

The reason I am offering this is because they are way faster and with multiple cores, only the interrupt for the core in use gets interrupts off, and because of the extra cpu speed means less time with them off.

The War1B have been a great client but I think they are maybe not the most suitable for an AP. I am reducing the price of the Ventana to help people at least try them as an AP.
Reply
#4
I am currently working on 9239, which is in the beta folder. Would you please upgrade to that and let me know if the OLSR is still an issue?
Reply


Possibly Related Threads...
Thread Author Replies Views Last Post
  New firmware release policy. lonnie 1 3,887 01-26-2011, 08:52 PM
Last Post: tony

Forum Jump:


Users browsing this thread: 1 Guest(s)