Supermicro PDB / PSU ''trouble with tribbles''

[azio]

Hey guys,

Been a while since my last post. Long time FreeNAS user here. So, had my 45 bay JBOD Supermicro 847 goin a while until about a year ago it broke before I flew to Seoul on business.

The story is simple. It's always had a red light and warning buzzer because something seemed to always be up with one of the PDB slots. As you may know the SM 847 has redundant psu. I have a pair of 1400w golds in there. It ran for a long time with the buzzer with both PSU in the PDB, one amber one green.

After, me, seemingly very mistakenly thinking it was a bad PSU, I replaced with a (yeah I know this was bad) a non exact matching PSU (tho the part number was the same), and I have great suspicions that the ebayer that sold me it may have sold me a bad PSU that has fried the PDB in some way.

Now when both PSU in the PDB I get two amber light, both PSU power up and self test OK amazingly, but the PDB doesn't appear to watch to switch either one of them. Now it may well just be coincidental, or the PDB got damaged more somehow when reinstalling/reinserting the new power supply I bought from ebay, right. But I have strong suspicions that the PDB has gone bad.

I'm trying to replace it, the thing is, it's a real donkeys mare to do it, and it seems like I'm missing good documentation, and couldn't find proper instructions in the manual for my specific model. I was wondering if anyone might be able to help guide me what the best way is to replace the PDB, and whether they agree with me about my diagnosis of the unfortunate condition my JBOD suffers from. I have a few arrays I'd like to get off this, and for least complications and simplicity it would be great to get my favourite disk array JBOD supermicro chassis back in action.

IF anyone has any tips about how I better I can troubleshoot the issue before replacing the PDB that would be awesome too. (please see images with comments attached).

At one point I was wondering if it was the power button, because literally everything seems dead when I press power button. Then again I'm not sure how to test that on the board, or how to replace the power button. I've tried to trace the wire and looked thru the manual for the SM 847 religiously, there just doesn't seem to be any detail about it. Sorry if the post is a bit long and stupid. It's been a while since I maintain server equipment like this. I think for the most part I'm following the correct routine, I just struggle to actually remove the PDB to try and test the replacement.

Is there anything else I could try before I replace it anyone else can think of that I may not have tried yet?

Thanks guys, I appreciate any ideas to fix my trouble with tribbles!


Best wishes,
A

 

[azio]

It's this model by the way https://www.supermicro.com/en/products/chassis/4U/847/SC847E26-RJBOD1 with the goldfinger PSU's

 

[azio]

Hmm guys. Looking carefully at the PDB and the specification for the power supply I have a feeling that I can see what has happened here.

It looks like this is the cause. Hopefully only the PDB is trashed. Is there any experts out there that can consult about whether my assertion is correct (See attached picture of the existing PDB).

I have another 847-8824 here I can put in. It's just proving a pain to find proper documentation or instruction to get to it. It looks like I have to take most of the fans out just to replace it, but I find that hard to believe..?

Thanks anyone who can help or provide this engineer a little direction, it would be appreciated. :-D

This FreeNAS JBOD has been dead for far too long and it is not prospering. Question really is if I'm right about this fault whether the PDB is replaceable or reclaimable. I have the replacement here anyways if need be. I just hope the rest of the board isnt trashed.

Best,
Adam

 

[joeschmuck]

I can't speak to how to remove or if your theories are correct but i can tell you how to remove solder.

WARNING - Do Not Leave The Soldering Iron on the board for more than a minute. Overheating the PCB will likely damage it.
1) Buy a small roll of Solder Wick (wire mesh with flux)
2) Buy a 15 to 25 Watt soldering iron or an adjustable soldering iron (more isn't always better and 5 watts might be too small for the job at hand)
3) Unspool a section of Solder Wick and place on top of the solder you want to remove.
4) Place the heated (let preheat about 5 to 10 minutes) solder iron on top of the solder wick and apply firm (not too hard) pressure. The heat will transfer through the solder wick melting the solder and the solder will flow into the hot section of the solder wick.
5) Inspect the area, if more solder is to be removed then place a clean section of solder wick over the solder and repeat.

I just located several soldering stations on Amazon that are adjustable 40-Watt units so you could adjust the heat needed just enough to melt the solder. $33 bucks is a very fair price. It's not a high end solder station but it would give you use in the future for different soldering applications, like a car, your Walkman (yes I'm dating myself), a computer power supply, adding LEDs. Sorry, can't do copper water pipes.

Good luck if you go down this path and take your time. If you need to reverse the process and add th solder back then you should purchase a small roll of silver solder, make sure it's thin for PCB work and not super thick.

 

[azio]

Hey Joe,

Thanks a lot for your reply and the kind tips for soldering. I should have probably mentioned I'm an half competent engineer and have a history maintaining units in datacentres and repairing hardware on occasion like this.

The problem is I can't even get to the PDB to remove it, if I wanted to, apologies if I didn't illustrate this well. I've had another go today and removed most of the fans and frame and it seems that the PDB is on a horizontal sliding fixture, which means (or seems) the locking pin I was talking about needs to be removed 90 degrees to it where it mounts up against the two power supply chassis enclosure.

This sucks and is a really terrible design of supermicro (as is the fact that the PDB has two pins soldered, this seems to be by design (or by oops I'd call it from a manufacturing error they've rush-fixed) rather than something resolved on the IC of the PDB. It seems solder has leaked onto the 3rd pin so hopefully I can repair or replace once I actually get to the unit!!

So the two main ports for the PDB literally have connections soldered together on the power ports. This seems like madness, but what is real madness is that I've spent hours trying to figure out how to get to the PDB. I've removed half the frame trying to get to the PDB. I'm still no further getting to the PDB.

I've looked in the manual and it doesn't really say anything about servicing or removing the PDB. I presume that its not considered a user- servicable component or I'm looking in the wrong place. I'm looking at https://www.supermicro.com/manuals/chassis/4U/SC847J.pdf

please please if someone can help point me in a better direction for figuring out how to get it out of the case it would be wonderful. I'm about half losing my mind over here this evening :(

It's breathtaking that this problem exists with solder melting or even being on the pdb ports by default, typical mid-manufacturer 'oops' kind of thing, I believe that could have been solved a lot better. The real insult (or embarassment) to the injury is that I can't even get the PDB out to inspect it or replace it, and believe me it's a nightmare without having this additional problem to resolve :(

Thanks again for your support. Please if anyone can help link me to some documentation for this procedure it would be a real life saver!

Best,
A

 

[azio]

OK, so its definitely the PDB. :)) Tested with this psu I had lying around and front panel comes up green, as does the CB2 controller.

yay.

So I just need to figure out how to remove this darn PDB I guess, and try and fix these bad solder points, short of replacing with the spare I also have sitting here.

If anyone can help me figure out how to get the darn thing out the chassis I'd appreciate it. I know.. it's fatally embarassing I can't figure out how to remove. I had no idea it would be so difficult. My understanding was it was just the 3 screws, but that damn locking pin and maybe something else just won't give her up. :(

Thanks again :)

Best,
Adam

 

[joeschmuck]

So you cannot remove the two screws on top of the PDB, next to the "locking pin" thus removing the metal piece the locking pin engages with? I would assume you have tried that already. As for the metal rectangle going down the side of the chassis I can't tell if it's secured by spot welds but if not, it looks like you remove it with brute force, it looks to be snapped in place but I could be wrong. If it were in front of me I'm sure I'd be able to disassemble it without harming it.

Glad you were able to diagnose the problem with certainty. But can you order a replacement part even if you do get it out?

Good Luck!

 

[azio]

Hey Joe,

Thanks for your reply, I appreciate it chap. I do believe that the screws you refer to shouldn't be removed, for a few reasons, so I have not tried to remove those two adorning the top of the PDB, but I am willing to consider the possibility that I am mistaken here, if your right the distributor design is a truly god awful one, and it would make no engineering sense to me;

Those two screws do hold the entire PDB unit together, it consists of two pcb's stacked by four metal spacers, however the way it is secured is like nut and bolt, and even if I was to remove the screws, the components between the PCB have hard point connections, and from what I know of circuitry it's not there to be removed but to provide support strength as a spacer. I thought this before I bought the replacement part, but having purchased a matched 847-8824 replacement which I inspected quite closely before posting here in desperation, I believe at least I'm not imagining it. Most of my electronics experience is on arcade machines and smaller electronics, so I am not the best.

To add to the case the particular reseller I bought the OEM replacement part from even had the good mindedness to specifically epoxy those top screws confirming my previous suspicions doubly, these are not supposed to be removed. He likely had buyers remove them not realising and got fed up with returns of broken components he couldn't prove weren't his fault, or similar, right.

Added to this, the geometry of the PDB means that the slots for the power ports actually stick quite far back into that back enclosure, and as a result of the slot-mount being at 90 degrees to those power ports for the PDB, it seems very much so that even if one was to remove those screws and go to the effort of removing the hard points, which im not even sure is possible, it may still be impossible to remove the PDB without doing something about the metal cover/bay for the power supplies themselves. Now, it may be that there is some give for the PDB to move to 90 degrees, before being pulled at 180, and that is how the mechanism works, but I was really hoping someone might have done this before or be able to help find me a service manual describing the process. The whole thing is very unsettling to me. I'm either completely wrong in my assessments, or correct and just don't know the right procedure for removing the pdb correctly.

In short, the 90 degree fixture of the PDB (its on these little sliders) for the PDB, and the way the PDB power ports seem to jut out from the pdb far into the redundant power bays enclosure seems to be in a manner that necessitates the removing of something that I am missing. My understanding that normally on supermicro's the 'lid' of the power bay is removed, by top screw on that, to necessitate this maintenance but there is no screw for the power bay on the 847.

Man I'm bust.

At this point the whole episode is embarassing, and I agree I could use force like you suggest and thought of the same here too. I actually believe the PDB in situe might be repairable, or would resort to brute force at this point, also these enclosures are still not inexpensive, and I've not done a PDB maintenance on a supermicro before. Never before has such a replacement been so manual, overbearing and mysterious or documentation been so lacking or poor. I know supermicro are not known well for their support, but this embarassment and mystery either reflects on my competence of an engineer, or the inability for supermicro to make a maintainable power unit. I hope it's the latter at this stage

Thanks again. If anyone has any ideas, please help!!

Best wishes,
azio

 

[joeschmuck]

Well you could buy a replacement chassis and move everything into it and call it a day. Then you could pull out all the screws and use some force to pop things open on the old chassis, but no hammers!

Yes, I'd remove those two screws. Hold the hex post with a pair of needle nose type pliers or a small open end wrench and take out the screws. I don't think you have anything to lose at this point in time.

 

[azio]

hey everyone,

Joe, thanks for the support and advice.

I'm happy to report that I've replaced the PDB. It was a 416-8824. I replaced with an 847-8824. I've checked the 847-8824 is compatible with my power supplie model PWS-1K62P-1R (they are gold 1400 watt goldfinger type). See (https://www.supermicro.com/support/faqs/faq.cfm?faq=24639).

Connected the 24 pin ATX as usual and loaded the power supply in. Plugged in, gave power. No green light at the front of the chassis. What a bummer. Exactly the same symptoms as with the other PDB. What gives.

I thought maybe I missed something here, plugged in a normal ATX power supply I have, greenlit the front of the chassis straight away.

For some reason or another neither the original PDB or the replacement pdb is giving me the green standby light at the front of the chassis. I just get both PSU reporting amber condition, and pushing the power button does nothing. When I use a regular ATX power supply I have which bypasses the PDB altogether I get green light at the front of the chassis straight away.

Does anyone know what's going on here? It seems unlikely both PDB's are bad. but we know that when I Connect regular ATX power supply directly bypassing the pdb i get a greenlight at the front of the chassis for the power button status, that indicates that it gets power normally, but isnt getting power from the pdb. Are both PDB's bad? Or do you think there is something different about powering the CSE-PTJBOD-CS2 board directly with a power supply that would give a green light that neither PDB I have here does.

It also seems strange that both power supply get amber and neither one goes green. So It's either both 847-8824 replacement pdb and 416-8824 original pdb are duds, i'm missing something or there is something wrong with the CSE-PTJBOD-CS2 board, but I'm confused as to what or how to test this further.

I suspect that there is a big clue given away in this diagnostic condition by the fact that when plugging in a regulat desktop psu to the ATX 24 pin plug and plugging in kettle lead, the green status light immediately appears on the front of the chassis. Whereas with the two pdb's and either single psu or both psu plugged in thru either PDB both will sit in amber, no 'greenlight' on the front of chassis indicating ready to switch on. This sucks!

Could the problem be with the way CSE-PTJBOD-CS2 instructs or manages the PDB? I see there is a connection between them, which means bypassing this is likely greenlighting the chassis. It stands to reason that there might not be a problem with the PDB and the cause is the CSe-PTJBOD-CS2.

If anything can please help with the best way to troubleshoot this it'd be super appreciated.


Best,
Adam

 

[joeschmuck]

Just some stupid questions to make sure I understand what you actually are doing and seeing...

1) With everything plugged in (both redundant power supplies) what is the status of the Power LED on the Power Board and Front Panel?
2) Now you depress the On/Off button, what is the status of the Power LED on the Power Board and Front Panel?

I suspect that one of the voltages is not present on the connector feeding the CSE-PTJBOD-CS2 board.
1) Do a Google Search for "ATX Connector Pinout", print it out.
2) Inspect the ATX power connector very carefully, ensure there are no pushed pins (all pins seated fully in the connectors).
3) Using a multimeter verify the voltages are accurate.
4) If you identify a missing voltage you will need to trace the issue back to the power supply. I'd only keep a single power supply installed and trace the suspect voltage.
5) Post your results.
6) Good Luck!

 

[azio]

Hi Joe, everyone.

Thanks again for the help.

1) The status of both power supplies is amber, indicating that the power supply are not coming up. I'm trying to find if this is intrinsic of a PDB failure or a CB2 power board controller failure.

Having got through the 3rd PDB power supply replacement which is an EXACT model replacement, pulled from a known working system, I confirm that PDB must be working correct. Likely all 3 I have are working correctly.

The feature of the problem is that I can turn the machine on, and indeed it turns its self on automatically when the unit is connected with a regular ATX power supply. However, when it is connected via either of the 3 PDB's I have nothing happens.

This leads me to believe that what is happening between the CB2 and the PDB is more complicated than the regular desktop power supply which will dump voltage into the control CB2 controller powering on the unit no matter what, whereas I know with the PDB it is more complicated.

2) Pressing the on and off button works perfectly with a desktop power supply, it also powers on automatically as expected when power runs. Unfortunately when using either of the 3 PDB's including the original I thought was faulty, nothing happens whatsoever.

It's soooooo frustrating.

It must be the CB2 right, I've replaced everything else apart from the 2 PWS-1K62P-1R units. I'm getting an Amber psu no matter what I do. But with a desktop power supply works right off the bat.

Given at this stage a regular desktop ATX connector works okay, and I've tried 3 different PDB, so different 3 ATX connector I think this is fairly thorough analysis from an engineering standpoint. I think my only last step is to replace the CB2 and hope for the best, because if it is not that. It must be the Power supplies themselves.

Though my understanding was when they go into amber like that, its the PDB thats holding them in standby mode. So either 3 PDB's , 2 replacements and 1 known working pull which I had imported from america at great expense are busted :( or it must be the CB2 unit itself that is faulting and giving a bad code to the PDB, making it halt.

I can't explain why else the PTJBOD-CB2 unit powers up fine with a desktop power supply but neither of my 3 Supermicro PDB's and power supplies. At this stage it's driving me nuts, It's not because I'm not a fairly competent technician or electronics engineer, it's simply because the problem isn't defined anywhere, i can't find a good specification or documentation of the protocol, there is no documentation for what is going on between PDB and the CB2 and it is likely some kind of proprietary mode that only supermicro use and it being a sort of specialist subject it's really , really hard to find information about this mystery.

If you or anyone can help clarify whether I am correct in my understanding or elaborate more about how power is controlled by the CB2 and PDB that would be useful.

I just tested the pin outs of the ATX from the PDB and I can measure 3v and 5v on two of the pins between the common ground, but I don't understand the power specification of ATX super well. I know that there will be 12V as well as many other voltages seen once it enter in the power on state.

This mystery might be untangable somehow if we can account for why the CB2 unit has no trouble coming on at all via a regular desktop power supply, but struggles with no less than 3 different PDB's.

I know at this stage, because I've ruled everything else out it must be the CB2 or the 2 PWS-1K62P-1R power units I use in the 3 PDB's Ive tried so far. Why on earth it works with desktop power supply really bothers and annoys me - it is some proprietary crap I think between the CB2 and the PDB I just can't make sense of it.

Please any help from anyone would be highly gratifying at this stage. Thanks again!


Best wishes,
Adam

 

[joeschmuck]

Just to reiterate, you plug a normal desktop power supply into the CB2 ATX connector and the system operates normally. If that is true then I suspect the cabling between the PDB and the CB2. Again this is where you need a volt meter and the pinout of an ATX connector.

I've attached a plug diagram and also reference this webpage for ATX pinout help. With power (115VAC) connected, ATX connector unplugged from CB2, take a voltage reading across pin 9 (power good 5V Standby) and and ground wire. You should read 5VDC. I suspect you do not have this 5VDC signal, it is the only power which should be on this connector at this time, well except for Pin 16 Power On which is pulled to 5VDC and you momentarily jumper this pin to ground in order to switch the power supply on or off. If the power is there as the ATX specs say then I have no idea what is causing the problem. But do inspect the connector pins carefully when you are reconnecting the connector to the CB2 board, make sure none of the pins get pushed out of the connector shell.

Do not mess with any other pins on the ATX connector when the unit is connected to wall power. They are not hazardous but you could damage something easily if you are not careful.

Edit: Corrected name of pin from power good to 5V Standby, it is still pin 9.

 

[azio]

Hi Joe,

Thanks a lot for your reply and the sanity you could give describing the ATX power supply specification voltage outs for pin 9, I wasn't intimately aware of PWR_good but fully recognise it's usefulness in the context of my situation. Thank you so much.

Yes, I can report that there is as you predicted no power coming from the PDB on pin 9. Goodness me.

This is the 3rd PDB though, including the original, first and the second replacement. I'm fairly sure at this point, we can be certain that one of the PDB's worked, and I am suspecting at this stage all of them worked, but the exact same behavior has endured of system being dead with each, and the only thing that endured during the problems (or was bypassed using the Desktop ATX) was the PDB and power supplies within it.

Am I to presume that it could very well be that both my PWS-1K62P-1R power supplies may have gone and not just the PDB, and I was too naive to consider that both power supplies could fail simultaneously? I honestly thought that was too unlikely and that amber was a good sign of the PSU waiting for the PDB? Afterall there is some voltages on the ATX pins. But maybe you know more about what's going on here than me?

I'm left to consider the logic following that if PDB is not reporting PWR_Good, that is because PWS-1K62P-1R is not good, I have 3 bad PDB's, or something is bad with the CB2 connector or circuit which led the pdb to put itself into a fault condition and therefore there is no signal on ATX PWR_Good. I think occams razor can rule out having 3 bad PDB's quickly, assuming at least one of them worked, and knowing that the desktop ATX power supply works and that PWR_Good is most likely not reporting on any of the PDB's I have here, at least with the two power supplies i'm using with the 2 replacement PDB's - one is an exact model replacement, or it is supposed to be so if the PSU's worked and the CB2 board isn't broke too, we should be up.


This hypothesis seems sound, but what do you think Joe?

Can it be possible that I can have a) either 3 working PDB's and 2 broken PWS-1K62P-1R 's or b) just one broken PTJBOD-CS2 (which even this tends to be disproven by desktop atx psu powering system). I think that it must a, maybe b, as c) 3 bad pdb's 2 of which from reputable suppliers seems too unlikely. I think the steps I should take is replace PWS-1K62P-1R first, then failing that the CB2. The only other small possibility is that I'm using a slightly newer version of the PDB 416-8824 exact replacement which might not be compatible with my existing power supplies. They seem to fit in just fine though and as I mentioned some of the ATX pins do get power on ground, so I am satisfied. The whole supermicro power conventions are a mess (indeed, it is a case nearly of what conventions).

Best,
A

 

[azio]

and yep, it works fine with desktop ATX power supply which is odd and really perplexing me having tried so many different PDB's im led to believe maybe it the power supplies inside the PDB.. it's just there are two so doubted both would fail, it seems like its either that or I am really extra ordinarily unlucky to get 3 bad PDB. or its some kind of magic in the connection between the pdb and the CB2

 

[azio]

I think there is a better way of summarising the above. So, does this mean the system powering on and off okay with desktop power supply indicates the error is certainly with the pdb and/or psu only?

Most importantly does psu good signal rely solely on how psu reports to pdb or can an error in the cb2 cause the pdb to fault and never enter a psu is good signal state? I don't know if there are any dependencies for the CB2 to start which the desktop ATX power supply naturally bypasses that also need important consideration to properly troubleshoot the issue. I have been working from the presumption that the 2 PSU I have are good and the PDB was bad but I have replaced 3 now.

That is the part I don't quite fathom.

In both cases it must stands to great probability that the PDB did not fail and that for some reason both power supplies did.

Now here is the thing, I actually have 3 PWS-1K41P-1R power supplies.

If I recall, the circumstances of the failure that brought this about it was beeping for a long time and possibly had a red light inside it, I believe flashing, which prompted me to buy a 3rd PWS-1K41P-1R power supply on ebay before I went abroad last year but when I did the maintenance to replace install the 3rd power supply I recall it stopping powering up altogether. Previous to that incident I had known for some time full well the unit would operate only with one power supply, which is what prompted me to seek to replace it, but I might have made a poor assumption that I think persisted til now that the PDB must have been on the way out when simply the second PSU was on the way out - maddening if so.

Evidently having found that 3rd PSU I had got back then just before that happened and tried to power it up I find it to be a brick - compelling. If either of the psu's or PDB's went bad, one thing we can be certain about is that this might be related. It is possible that I got entirely unlucky or my PDB fried all 3 of my PSU somehow..When my new PSU arrive I will have replaced the pdb and the psu, leaving only the CB2. That's pretty much everything then. The sadness will be if i replace the latter and still can't figure out what's going here between the CB2 and the PDB. It can't be all 3 pdb or psu that are bad surely, you'd think.

So, maybe it was just a bad psu and pdb that bust them all, and retrospectively I should have replaced both psu's and then the pdb before I even thought about replacing the PDB twice.

I think this is a lesson learned. I should probably know better than this. The plan is to order a new pair of PSU for the fresh PDB and hope to god the problem was simpler than I had first anticipated!! In fairness, besides replacing the backplanes I've replaced nearly everything else trying to get to the bottom of it!!

Best,
Adam

 

[joeschmuck]

I have no idea how well built/designed the PDB is built but a reasonable person would think it would be resistant to power issues unless it failed and caused the PSU to fail in the first place. But also in the future if you have a PSU failure you could just plug in a single PSU and see if that works flawlessly and test each PSU for functionality. Once you are sure all is good then you can have a pair plugged in.

 

[azio]

Joe, totally.

Having had the original 2x PWS-1K41P-1R Power supplies and none working in redundant or single mode, and a replacement PWS-1K41P-1R that arrived last year was not working either led me to this.

I made the false assumption that the PDB was bad, when it might have been a good one. Since it is a common issue with the chassis as they are getting quite old. It looks like I might just have 3 bad PSU's after our talk and a deeper look, that seems way more likely than 3 bad PDB, two of which are new (and certified /tested by the seller and supermicro etc).

I think I was getting confused a lot by the way the CB2 and the PDB interacted too, and baffled why the chassis worked fine with an independent ATX power supply but not with ANY of my PDB's - likely because of PSU - right. I feel dumb now, if this is the case. At least I will feel better if the problem persists when a new PSU arrives tomorrow - but I feel worse if it doesn't for being too stupid to consider something nearly equally as possible, given the maintenance was for power failure after all.

I have a fourth tested PWS-1K41P-1R unit shipping to me tomorrow from a reputable supplier. Hopefully it will work. I think if it doesn't I'm missing something obvious, I do refuse to believe I have 3 broken PDB's, 2 from different re-suppliers. Given that it was also plausible that only 1 power supply was known working with the original PDB. It is plausible that the 3rd PSU purchase was a dud and the 2nd had already failed, not the PDB, a marvelous coincidence, but not at all a completely unlikely one outside of the area of reasonable possibility.

My faith in logic will be restored if the new psu works, as someone who is usually able to resolve these sorts of issues quickly, I am shaken that its taken me weeks to do such a simple maintenance (theoretically a year as I was abroad!!) and it was simply through a reluctance to replace both the PSU and the PDB, the only way to be sure. The danger of an assumption in troubleshooting. So I only have myself to blame and a slight potential overzealous willingness to believe the situation might be worse than it is. But who knows maybe the PDB and the PSU's were broken, but the step I missed was getting another known working PSU as well as the PDB, I see that now. 1 or 2 out of 3 might seem like good odds, but ultimately I was gambling my psu's were good. I think we've got this.

I'll just be glad to know I have a fresh PDB and working PSU's in it and have my arrays back if this works!

God, please work :D lol. I have to be so unlucky if it doesn't.

A

 

[azio]

The new PWS-1K41P-1R PSU unit arrived this morning.

As soon as I put it in and plugged in for a power on test. It powered right on up first boot.

Damn I felt dumb. I don't yet know for certain that the first PDB is definitely dead, maybe that is what prompted not 1, 2 but 3 other PWS-1K41P-1R psu to fail. It is entirely conceivable that either the first PDB was the cause or I was unlucky enough to have 3 PWS-1K41P-1R failures in a row, and one duff PWS-1K41P-1R replacement, that I might have only ever had 1 working one.

My reluctance to consider that caused much time to be spent, but on the bright side I am glad, as I have now become somewhat expertly familiar with this chassis and a fine one it is.

But damn I feel like a dumbass not trying to replace the PSU a fourth time before replacing the PDB twice, considering how much easier that is.

All is well that ends well. I fully reassembled the chassis and all my arrays running as usual. Maybe others can learn from my experience.

Thanks for the help, it was meaningful. And greatly appreciated!

A

 

[joeschmuck]

Glad you FINALLY found the problem. I hope it stops giving you grief. Take care.