Thankfully I had been alerted to the "2.0" compat flag issue before updating and had already set it via uci commands, so that doesn't seem to be what caused my bricking. (And no, it wasn't a forced update when mine bricked.)
Errrmmm, what?!
Blindly overriding the compat flag and totally ignoring its reason would be exactly your issue. The flag itself is meaningless, the reason why it has been bumped however is not, far from it. So you removed the "STOP" sign - and drove happily into the flowing traffic, after all there was no sign anymore...
1 Like
thanks for the tip. unfortunately:
Using ethernet@1b100000 device
TFTP from server 192.168.1.254; our IP address is 192.168.1.1
Filename 'openwrt-mediatek-mt7622-linksys_e8450-ubi-bl31-uboot.fip'.
Load address: 0x48000000
Loading: #################################################################
#####
3.7 MiB/s
done
Bytes transferred = 1019796 (f8f94 hex)
"spi-nand0" partitions still in use, can't delete them
MTD device fip not found, ret -19
Press ENTER to return to menu
and...
Using ethernet@1b100000 device
TFTP from server 192.168.1.254; our IP address is 192.168.1.1
Filename 'openwrt-mediatek-mt7622-linksys_e8450-ubi-preloader.bin'.
Load address: 0x48000000
Loading: ######
1.8 MiB/s
done
Bytes transferred = 88077 (1580d hex)
"spi-nand0" partitions still in use, can't delete them
Erasing 0x00000000 ... 0x0007ffff (4 eraseblock(s))
"spi-nand0" partitions still in use, can't delete them
Writing 131072 byte(s) (64 page(s)) at offset 0x00000000
"spi-nand0" partitions still in use, can't delete them
Writing 131072 byte(s) (64 page(s)) at offset 0x00020000
"spi-nand0" partitions still in use, can't delete them
Writing 131072 byte(s) (64 page(s)) at offset 0x00040000
"spi-nand0" partitions still in use, can't delete them
Writing 131072 byte(s) (64 page(s)) at offset 0x00060000
Press ENTER to return to menu
1 Like
IMHO, it’s quite unlikely that power supply will only cause flash memory corruption. It should cause random router crashes too if we have questionable power supply.
The OKD issue looks like major software bug to me.
Maybe I’m lucky? My 3 E8450s has been running great with my own master builds. All 3 are still on 5.15 kernel and installer v1.0.2 or earlier, with H/W offload and WED enabled.
2 Likes
I have three Belkin RT3200, all of them have been upgraded to kernel 6.1 / FIP-in-UBI setup using @daniel's v1.1.0 installer. All the three devices were bought in USA and use the original Belkin provided power supply.
Two of the RT3200 are powered on right now. The 1st RT3200 (I will call it RT3200-1) does not exhibit any issues (as of now). The 2nd RT3200 (I will call it RT3200-2) occasionally does not power ON after a reboot or sysupgrade. But power cycling usually turns the RT3200-2 back ON.
The 3rd RT3200 (I will call it RT3200-3) was left powered off for days, and yesterday when I connected the OEM power adapter it did not turn ON. After reading through some posts in this topic that suggested keeping the device in freezer, I left it in freezer for about 3 hours and after that the device powered ON.
I currently do not have the tools to connect Serial or JTAG in order to check what is really going on on either RT3200-2 or RT3200-3.
EDIT: Sigh. Now, after trying to sysupgrade RT3200-3 to MAIN SNAPSHOT r25580-85ad6b9569 (Image Builder built, not auc), it did not come back ON after the reboot triggered by sysupgrade. Back into the freezer it goes.
look here, I had already updated when the attended sysupgrade later failed and bricked my device.
When you say you had already updated, after you ran the latest installer and had the router running on an initramfs image in memory, did you follow up and sysupgrade to write the permanent 6.1 image to flash? I'm guessing you probably did, but failure to do this seems to be causing a lot of problems.
FWIW, I don't trust attended sysupgrade. Firmware selector builds only. Call me superstitious, but I've had no problems with firmware selector images. I might trust attended sysupgrade for a point update within a stable release, but that's about it.
1 Like
Has anyone replaced the antennas on their RT3200?
I'm reorganising my office soon, and my main desk will be right at the ethernet port - problem is, I can't fit the RT3200 on the desk, so I was thinking about 3D printing an enclosure, with an added fan (via the USB port) to reduce its profile. My only issue is the antennas, don't want to mess around with the glue much.
Some recommendations (especially ones available in Europe/UK) would be most welcome!
Will the forthcoming new 23.05 release be safe to install without running the new installer?
1 Like
Did RT3200-3 resurrected after the second "ice age"? I switched off my semi-bricked dumb AP RT-2 and I hope after some days in the shef it will boot again.
I've done exactly that on not only my bench unit, but also on one of the regular use units as well. There's more than enough space for terminating a couple U.FL to RP-SMA connectors at the top of the back panel. This would give you two external antennas and two internal. It's also possible to place a couple on the top of the case as it is normally oriented, but this requires modifying the sled and removing the area where the front pair of antennas normally go.
From there, the only worries you might have would be whether or not you're violating regulations on transmission power or type certification. Any modification of the antennas will also change its radiation properties, so you could find the device is radiating more or less power than it thinks it is.
2 Likes
I will be printing a fully custom case, so antenna placement isn't an issue 
But basically then, any 2.4/5GHz tuned antenna would work with those adapters?
Any 2.4+5GHz-tuned antenna with an RP-SMA connector would work with the adapters I purchased. There are some antennas out there with standard SMA connectors, so buyer beware. As long as you have a U.FL for connection to the board, the antenna-side connector is entirely up to your choice of antennas. As always, try to keep the cables as short as possible. Most of these thin cables will absolutely destroy the signal at any real length.
1 Like
Yes, 23.05.x will stay on the old installer and upcoming 23.05.3 (and all further 23.05.x releases) will not require running the new installer.
5 Likes
Maybe I missed the commit - could we please get the performance governor set as new default for all mt7622 targets in 23.05 branch before 23.05.3 is released? To minimize reboot bricks running further 23.05 stable series and follow MediaTek vendor defaults.
1 Like
Try using a 6-pin JST connector on the router serial interface. I was trying to make a connection without that and had symptoms similar to yours.
1 Like
Thanks! I'll try to get one of those connectors hooked up and see if that helps. Seems odd that the little pin connectors that I've been using seem to work fine with other OS's using the same USB serial device, but I've seen stranger things.
Yes, RT3200-3 came back to life after a short stay in freezer. After some time outside, a reboot put it back in non-responsive state. Again it powered on after a stay in freezer. It is rebooting fine if I trigger the reboot when the device is cool enough. After being outside and reaching room temperature, it doesn't come back on after a reboot.
I tried this about 3 times (using the OEM power adapter). All 3 times it turned on when taken just outside the freezer but fails to reboot after reaching room temperature.
I am guessing when it is cooler, the internal electrical resistance is less and current flow through the device is sufficient enough to turn it on but at warmer temperatures, due to higher internal electrical resistance, the current flow drops which is probably not enough to power on the device.
OR
My above theory may be completely wrong. Just thinking out loud.
Also not sure if the supplied OEM Power Adapter (Output 12 Volt, Max 2 Amp) is causing any issues.
I also used the power adapter of my GL.iNet GL-MT6000 (Flint 2) which uses the same DC Barrel Jack as the one in RT3200 but this power adapter has a higher output rating (12 Volt, Max 4 Amp), but the RT3200-3 still did not power ON (tried when RT3200-3 was at room temperature).
1 Like
I just bought a used RT3200 and installed the prebuilt v1.0.3 UBI release version which I believe installed OpenWRT 23.05.0. I just spent a couple of hours scanning through this thread discussing this issue over the past few months after I stumbled upon a post about this on Reddit.
I've set my the CPU governor to performance and noticed that the temperature increase is a bit more than the reported ~2C bump. I believe mine have increased at least 3C.
For the past couple of days, I've noticed thermals idling around 57C and fluctuate upwards of about 58-59C at "moderate" loads. Before applying the governor change, thermals were a bit below 58C. About 45 mins after applying the change, it seems to have stabilized at around 61.5C. Note, the temperature spike at the beginning of the graph is due to iperf testing of the 5ghz performance.
First question: will these thermals cause long term damage? I thought I saw someone post here that theirs were about 10C lower in the upper 40s to lower 50s range.
Edit: House ambient is 68F, about 20C. The device is in a closet. So it's perhaps a a degree or two higher where the router is located.
Edit 2: I've installed SQM, Adguard Home, vnstat2, and luci-app-statistics. I also have a USB drive plugged in and written to by the system log, adguard, and vnstat2.
Second question: would you all recommend upgrading to a newer build?
The only issues I've come across:
-
A suspected misconfiguration of the firewall after installation of Adguard Home caused the system to become fully unresponsive. The wireless networks were still visible. But I don't believe any devices could connect. Nor was there any internet connectivity. Luci was unresponsive, SSH not working. After rebooting the device, connectivity was restored with the device seemingly functioning normally. A bit more configuration tweaking to the firewall and Adguard Home was working to my satisfaction. For those wondering, I believe I was trying to restore DNS lookups to the guest network after DNS seemed to be broken after installing Adguard Home.
-
I didn't realize it at the time. But on a separate occasion, I might have had an instance of the BL2 DRAM calibration issue encountered by many. I believe was making some changes and initiated a reboot via Luci. After waiting an unusually long amount of time for connection to be restored, I noticed that no lights on the device were on. I flipped the switches on and off a couple of times. I pulled the power, gave the power switch another couple of tries before it worked. I've rebooted at least a couple times more since this occurred without issue. I had the CPU governor and min frequency set to the defaults.
I'm unsure if this is typical or not. But it seems like it takes several seconds after powering on the device before any LEDs turn on. It's definitely more than a couple. Probably less than five. And I'm not sure if it's consistent.
1 Like
I had one more question about 5ghz band performance. I can get line of sight speed of ~550 Mbps. This is at about 10 ft away using an iPad while standing motionless. But packet loss is approximately 1-2% when I check
iw dev XXX station dump
This is after a 30 second iperf test. Retries are always equal to failed packets in the "iw" command output.
Putting a single interior wall with 1/2" drywall increases packet loss to over 3%. From regular use about 20 ft (edit: actually about 28 ft) away and perhaps two walls (edit: four walls due to needing to go around another corner) in between, I've seen packet loss over 10%. RSSI in this case is in the mid -60s dBmV as measured using Airport Utility from the iPad.
Am I wrong in thinking that this is a bit high amount of packet loss?