RR4360X Power Supply in enclosure internal is no longer detected

[m0rthy]

Out of the blue, our RR4360X started reporting the PSU are no longer detected, while they are in and working thankfully

Tried to unplug and plug to refresh state - no result

Tried to change PSU slot - no result

Has anyone faced that and have some remedy?
Attached are logs and screen

[Sandshark]

By "out of the blue", I assume you know for sure nobody opened up the unit.  The power supplies are monitored via the I2C bus, and though it's capable of a lot more, may not be used for any other devices by the ReadyNAS.  There should be a (usually 4-pin, but I've never had a 4360 to look at) from the module the supplies plug into to the mother board.  If that got knocked loose, that would be an issue.

Also, no firmware updates?  There is a section of ./etc/sensors.d/system.conf (which is itself a link to ./etc/frontview/sensors/RR4360.conf) that should look like this:

chip "pmbus-i2c-*-*"
        ignore in2
        ignore in3
        ignore curr1
        ignore curr2
        ignore power1
        ignore power2
        ignore fan1
        ignore fan2
        ignore temp1
        ignore temp2
        ignore temp3

chip "pmbus-i2c-*-58"
        label in1 "PSU 1"

chip "pmbus-i2c-*-59"
        label in1 "PSU 2"

chip "pmbus-i2c-*-5a"
        label in1 "PSU 3"

chip "pmbus-i2c-*-5b"
        label in1 "PSU 4"

It's possible Netgear forgot something for the 4360 in a firmware update, though I think we'd likely have heard about it from other users before now.

The last thing is that one supply can cause a problem on the bus for all, But in your swapping, I assume you checked that the problem wasn't fixed with any one of the supplies removed, so that's not it.

Unfortunately, since it's a complex bus and not simple discretes, troubleshooting will be difficult.  The I2C hardware is typically in the supplies themselves and, of course, the motherboard.  The module into which the supplies connect usually only serves as a means of connecting the separate supply busses together (just "wired" together, no active hardware).  So unless it's one of the possibilities I listed above, the problem is likely the I2C hardware on the motherboard itself since all four supplies are affected.  If they made it more complex than necessary, there could be a multiplexer in the PS interconnect module, which could go bad.  That module is likely a standard Chenbro one, as Chenbro appears to the the real designer/manufacturer of the OS6 based rack mount systems.

There is a program called i2c_tools that may help.  Do the following via SSH:

apt-get install i2c-tools
i2cdetect -l

If that detects the I2C bus controller, then that's a good thing.  If it also detects the power supplies (bus addresses 58 - 5B per the configuration file), then it's a firmware issue.  Unfortunately, that's the extent of what I know about I2C bus.  But if you post the results of the i2cdetect -l command, I'll try to help interpret them, and maybe somebody else has more expertise in this area.

[Sandshark]

I did think of one more thing the PS interconnect unit might do:  set the power supplies' bus addresses.  Something has to give them separate addresses, and it's likely jumpers in the interconnect unit.  If they are pull-ups/downs rather than shorts, a common resistor pack could be to blame.

But let's see what I2C-tools has to say before drawing any conclusions.