AintBigAintClever
HPE Aruba Networks have removed all older firmware for some of their high–end ArubaOS-CX network switches from their website, because they prematurely wear out the switches’ onboard eMMC flash and cause them to fail.
Here are a few other failure modes of the 6300M series switches (the JL659A model at least) together with a means to recover at least one of them, whilst at the same time applying the all-important firmware upgrade that fixes the wear-out issue.
COMMANDS USED (once booted to ServiceOS console)
mount usb
ls /mnt/usb
update primary /mnt/usb/ArubaOS-CX_6400-6300_10_05_0021.swi
allow-unsafe-updates 30
boot
El Reg article about it: https://www.theregister.com/2020/11/09/aruba_6300_6400_switch_os_upgrade_urgent/
INDEX
0:10 Introduction: Urgent firmware upgrade for Aruba switches
0:28 Why the urgency? Oh, a bug that causes premature failure!
1:07 What it won’t fix: a dead switch with the “dot of doom”
1:25 What it won’t fix: a power supply that’s blown up
1:36 What it CAN fix: a switch stuck in a reboot loop
2:08 What you’ll need
2:35 Connecting to the USB console port
3:30 How to upgrade from USB
4:55 What caused the boot loop in the first place?
ARUBAOS-CX FIRMWARE VERSIONS
10.05.0021: FIXED
10.05.0020: defective
10.05.0011: defective
10.05.0001: defective
10.04.3031: FIXED
10.04.3000: defective
10.04.2000: defective
10.04.1000: defective
10.04.0040: defective
10.04.0030: defective
10.04.0020: defective
10.04.0010: defective
10.04.0001: defective
10.03.0110: defective
10.03.0090: defective
10.03.0083: defective
10.03.0081: defective
10.03.0080: defective
10.03.0071: defective
10.03.0060: defective
10.03.0050: defective
10.03.0040: defective
10.03.0031: defective
10.03.0020: defective
10.03.0010: defective
10.03.0001: defective
10.02.0060: defective
10.02.0050: defective
10.02.0040: defective
10.02.0031: defective
10.02.0030: defective
10.02.0020: defective
10.02.0012: defective
10.02.0010: defective
10.02.0001: defective
10.01.0050: defective
10.01.0048: defective
10.01.0040: defective
10.01.0030: defective
10.01.0020: defective
10.01.0011: defective
10.01.0001: defective
10.00.0019: defective
10.00.0018: defective
10.00.0016: defective
10.00.0015: defective
10.00.0014: defective
10.00.0013: defective
10.00.0012: defective
10.00.0010: defective
10.00.0008: defective
10.00.0007: defective
10.00.0006: defective
10.00.0005: defective
10.00.0004: defective
10.00.0003: defective
10.00.0002: defective
SWITCH RANGES AFFECTED
8400 series
8325 series
8320 series
6400 series
6300 series
6200 series
Oh no, we have a shitload of these Aruba switches on our premises! Dang.
Heh. Not the first time HP(e) have had run-ins with firmware being abusive to flash. iLO4 used to chew up its flash too. That was fun.
3¼mips…eggnchips®€cu™
https://youtu.be/zsjTpFR0oYQ
Lovely – just getting 16 6300s in School… I will make sure the RedCortex engineer updates as he goes. it seems that someone will be running around updating firmware in a hurry in the schools already victims of the "amazing" Aruba update project!
"Why did those switches fail to upgrade? Well, both switches, as it turns out, have ports which are failing a self test and none of our other switches had this. This switch is one of them, but now that it's been pulled out of service and allowed to cool down for a few days, it's cleared it. So we've now got a 12,000 pound switch that appears to be working, so we can't really RMA it. But do we dare put it back into production?"
NOPE! Absolutely not. Return to sender and get a new one anyway is what I'm thinking. Network switches that tear up their own SSDs/eMMC cards due to crap firmware, have PSUs that smoke themselves after barely five weeks under load and start failing self tests of any sort have clearly demonstrated themselves to be highly unreliable and are not to be trusted. I don't care how much it costs. You can't be having that stopping up business in these times. HPe need to get their shit together and soon or they're gonna be on the receiving end of lawsuits galore. Also, you can't be certain how long it will be before the other switches in the system start failing in this way too. I'd be keeping a watchful and suspicious eye on those switches if I were you.
Just to answer your final question on the video, yes please RMA the switch that has port failure only when warm. The switches are covered by the lifetime warranty.