I already searched for an answer inside the forum but I couldn’t find anything useful for my case.
Yesterday my CRS310 8G+ 2S+ died randomly, I take out of my rack and powered separately because restarting it didn’t help a lot. I connected to my computer thinking maybe netinstall could help. I plugged the power and no activity lights showing and I was noticing that the blue power LED was keeping flashing costantly, I waited 3, 10, 15 minutes but nothing was changing. This morning I retried thinking that some little fresh air inside the pcb could help but nothing changed. What a strange behavior, I’m astonished, I think that it could be my first and last 2,5gbe switch , I’m a student that built it’s own homelab with all my savings, sometimes giving up to other things and I’m a little bit disappointed.
I recorded the two tests: https://streamable.com/4h1o55 https://streamable.com/ff79iz
There is a common enough failure mode of power supplies when they do not actually stop working altogether, they continue to provide the right Volts but they cannot output enough Amperes, and since when booting there is normally a peak of current required by the device it may not boot.
Try describing what happens when you power it, compare with:
I tried 3 different 24v power supplies with an amperage higher than 1.5 amps, I cheked the voltage inside the board and it’s stable also during bootloop
I already seen the 3d and the difference is that he could netinstall while I can’t because the interface on ether1 doesn’t come up in my case. I kept holding the reset button also for 30 seconds but nothing changed
yeah, about two weeks ago the voltage regulator exploded, I replaced it but the fan wasn’t working again so I plugged a external PWM controller, the switch worked without problems for about 2 weeks, everything seemed normal until now. The board looks good without any burning mark or specific heat zones. It looks like a software issue
It looks like you know your way around electronics a bit but …
… what makes you think this is a software issue to begin with ?
Voltage regulator exploded, fan didn’t work anymore so that definitely counts as damage.
It worked again for about 2 weeks after some McGiver-intervention and now nothing anymore.
Please tell me why you think in the first place this is a SW issue ?
And what SW issue might that be ?
A module detecting power regulator has exploded and then decides to shut down the system because the original regulator is not present ?
It’s not because you don’t smell “the odor of frying electronics” nor you can’t spot heat zones, the board is not … damaged beyond functioning.
Side-note: and you thought it was not necessary to mention those little details in your first post ?
Oh, and you are disappointed because all of this ??
Unbelievable …
Because the board itself is good, it looks like it’s bricked or something.
Fan controller is not working, not CPU, ram or ethernet controller and before it was working flawlessly.
The switch think that fan isn’t plugged in because the connector on the board isn’t used but that wouldn’t prevent it working correctly.
We aren’t talking of a switch that costs around 30 euros, it’s an expensive model and it stopped working and I have a piece of malfunctioning metal in my hands. I’m speechless. I can’t find a way out to get it back functioning.
And what external factors caused that power regulator to explode ?
You don’t think it sent power spikes further down the line to whatever was behind it ?
I don’t know if you have realized it but you are responding in a completely unhelpful way without providing any real help, moreover going out of context with nonsensical and inappropriate sarcastic jokes.
you’re lucky i’m not gonna report it, I already feel so sorry for you as it is for your condition
I think you are misunderstanding, you have all my sympathy for the misadventure you went through, but sometimes devices fail, and and you have to accept it as a fact of life.
In a professional situation (with a suitable budget) anyone would have called the device dead when the fan controller exploded and - at the most - used the workaround of the external controller to keep things going until the new device would arrive.
Anecdotally it is a matter of luck, many years ago I got a hard disk that was thrown away because it had a (bad) head crash (a PATA Samsung 4.1 Gb), physically hitting the platters, I tested it, re-partitioned it in such a way that the defective areas were excluded from storage and put it as secondary backup on a machine, where it worked just fine for several years, you never know.
I wasn’t offended because it’s the reality, it sounded provocatory by you. Every student don’t earn money, I only keep money apart for reaching the budget and buy someting I need to add to my network rack, network appliances (the quality ones) are expensive and I bought it that was only open the box but never really used by the seller and it was working as new. Before throw it away or use it as replacement parts I was hoping that I could find a way to repair it and not buying a new one.
yeah I thought this option, but it would fry all other components also, not a specific zone (in this case the fan controller area)
I understand, but while what happened is unfortunate it is completely independent form your (financial) situation.
The little chip (or other component) that blew out or however failed to function properly doesn’t care about that, it simply blows because it feels like doing it, just like another component decides to keep going.
It is perfectly possible that after all the thingy is actually worikng but has something disabled/deleted, but if you don’t find a way to attempt netinstalling it, the game is over or nearly over.
There could still be a possibility (provided that the device is operational) to access the device (and if needed re-flash the bootloader) via TTL, it has been documented recently to solve a multiple RB5009 accidental bricking issue), but it is not at all easy-peasy, JFYI:
and not necessarily the same processes will be applicable to a CRS 310.
Not necessarily. Some components can handle such events better then others.
Some can survive the initial blow but will then gradually degrade… how fast, when, where ? Nobody knows. You can calculate and estimate probabilities but it’s never an exact science.
Which still leaves open the question: what external factors caused in that initial explosion ?
Self-inflicted ?
Did you at least isolate or remove the blown controller? Or just plopped in a PWM controller while exploded parts remain connected to circuits with highly sensitive parts like processor?
TBH without determining the cause of damage which you’d normally discover when systematically undoing any damaged sections you most likely kept the board operating with some fault that finally took out other parts of it.
It being on e-bay most likely it was already accidentally damaged - like blowing original PSU, connecting it to wrong voltage PSU, or prolonged exposure to reversed voltage and offloaded after realizing the problem or already observing some issues.
Now your best bet is starting from beginning - the fan circuit you know was bad and doing a proper repair. If you’re lucky and it isn’t some critical component that’s already gone you might restore it.
Even if the part failed on it’s own, it could’ve still taken out other components, by bridging incompatible voltage domains. Only by repairing the failed part you can discover and fix any subtle damage to surrounding components. How you know the fan is ok? Have you made sure it doesn’t feed 12V power into 3V3 RPM counter logic? Etc.
Also, when facing an issue, remove all extras, like SFP modules.
, I’m a student that built it’s own homelab with all my savings, sometimes giving up to other things and I’m a little bit disappointed.