Hi ...
I've been inspecting some RBs last two days, driven I was by a weird (or not) behaviour: a brand new 433AH started rebooting 3 or 4 times a day and after each reboot the bad blocks count increase a bit.
A couple of days ago this board became a production unit with V5.8, 1.2% bad blocks. Now after each reboot this percentage grows ... 1.6, 2.1, 2.4 and now it's 2.7.
I killed all "write to disk" threads I found: graphing, DHCP, etc in order to keep the disk "as is". I even disabled the autosupout.rif generation after each reboot. The period between random reboots are bigger now than it was when graphing & dhcp was writing to disk.
This board replaced a 433UAH that has 0.7% bad blocks but those random daily reboots. After formatting NAND via RS232 console a couple of times (and netinstalled 5.8 the same amount) the 0.7% dropped to zero. I exported the 433AH config and imported it to 433UAH (after correcting several export glitches 5.8 is generating).
This clone is at the bench now with 2 R52nM with the same MAC addresses from the production unit, with two clients running Btest (TCP) at solid 9M down / 4M up, 100% CPU for ... 12 hours.
Ok, on the real life this AP will handle maximum 30 + 30 CPEs but those reboots are not associated with auth clients count (nor with CPU load neither trafic load).
My bad blocks inventory: one 433 with 7.8%, another with 3.3%. A 433AH with 2.7% and counting. And the retired 433UAH now is 0% after several NAND formats + netinstall.
Question to MT guys:
1) Where can I buy NAND chpsets to be replaced locally (I have how to re-work, re-flow SMT devices).
2) The ROS licences are tied to NAND?
3) Is bootloader hosted by NAND?
Thanks.