Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[sevenlayermuddle]

Genuine incident, in the past hour.... It was my turn to make a trip to the kitchen to refill wine glasses.   In the hall, I smelled what I thought was rather nasty candles which was odd, as there were no candles burning.   On enquiry from better half, immediate reaction was 'I'm wearing that new perfume', is that it?' 

Cutting to the chase, the funny smell was coming from our main server.   It's a MythTV backend, plus  SVN source, plus bug database, plus media store and all sorts.   Specifically, the PSU is smelling very odd, I'll dismantle when I have time and see what's been cooking.

It's a decent box, HP Proliant, so I'd always hoped it was relatively safe to leave on.   Not so sure now, even with a new PSU.   

Only saving grace, we'd just finished watching Dr Who on Myth.   Else it would have been ever so tempting to leave it powered on til we caught the end.

[AArdvark]

I would not worry yet 
Power supplies do cook now and then.
Usually you find if you switch them off at this point, they do not come back or go 'Pop' on next start up. (Take care).
If nothing else has been also cooked you should be OK with a new *good* PSU.
If anything else plugged into your motherboard has also been hit, it is 'flip a coin and hope' time.
Regardless of how good the M/B is I have found that in these instances the M/B will be heading for something failing quite soon.
I am running a M/B that has a failed Gigabit Ethernet port due to the mains supply (not PSU) bouncing 5ish years ago .... it still works fine but I don't trust it. 
On the other hand I have had M/B fail because someone 'kicked' out the power cable [Kettle lead] by accident. (Should not be possible but it did happen  )
[Learnt not to plug cables in at ground level, even if only temporary, if it is possible someone could catch the cable. !!]
I do not have a UPS as it is a bit of a big spend for me, for proper coverage. 
(Most of the cheaper UPS's are rubbish as far as I am concerned. I have used good UPS kit in office environments and cannot bring myself to risk buying the rubbish Home users get sold.)

Best of luck with the investigation. 

[sevenlayermuddle]

Best of luck with the investigation. 

Thanks.   I'm actually out on travels next week and tomorrow the lawn needs cut, but I will post an update in time.

I'm feeling a bit smug lucky too, mind you.  The server snapshots itself (and other networked systems) every night.   Once in a blue moon (a couple times a year), conscience kicks in, and I burn the latest snapshot to a Blu Ray disk.   And oddly enough, I burned one yesterday.   I've no doubt the  hard disks are still OK but all the same, I better look after that Blu Ray.   

[AArdvark]

As you are SO lucky ...... and these things are supposed to come in three's, do you have any numbers for the Euromillions lottery ?

 

[sevenlayermuddle]

Not the PSU after all, smell was just coming from nearby Sata power cable.   

There should be an attachment showing a melted Sata connector, but not sure if I've attached it right?

Have to say, I've never felt entirely happy about these connectors, they carry a fair whack of peak current and yet are flimsy, and prone to getting accidentally knocked out of place.  That may be what happened here.  It's one of the originals fitted by HP at manufacture, so quality should be decent.

The disk itself looks OK, though I may well replace it even if it still works.  It was the volume that hosts TV recordings, hence the timing... incident occurred just after watching Dr Who. 

This thread may warrant being moved to 'PC hardware'?

[AArdvark]

That is unbelievable !!
I have never seen a burnt/melted sata power cable or come to think of it any power cable in a PC/Server etc.
Can I ask what drive type it was and how old ?
What else was connected to the same cable (I can see another white power connector in the pic)
If a short or something set the plastic on fire, you are lucky the drive still works.
Very odd & lucky at the same time

[sevenlayermuddle]

It was/is a 1TB Hitachi Deskstar, manufactured 2010.  They're generally regarded as decent disks, and my personal favourites.

The disk looks salvageable, the melted crud wipe off easily, leaving clean-looking  PCB tracks.   I'll probably want to connect it up one more time, if only to rescue some favourite TV recordings.    But I'll tread cautiously.  In dim & distant past  (as a kid) I used to repair radios and TVs for pocket money, and I remember that crud from component burn-ups tended to be more conductive than it looks, and very hard to remove entirely.

The connector is just a Molex to Sata splitter, like this...   http://www.amazon.co.uk/dp/B0066DA9N8

The connector's other limb was in use too, and is fine.   Mine is stamped 'Amphenol' and carries an HP part number.   No cost-cutting for me to feel guilty about, thankfully, other than maybe not making sure the connectors were fully home?   It's  a long time since I took the covers off that machine so even if badly connected, it had been working fine for ages.

[AArdvark]

1TB Hitachi Deskstar

Yes, good drives that just work 

Was not implying 'cheap' parts just aware that whatever was also connected to the other 'spur' of the cable might have been affected.
Modern drives are supposed to be able to take over-voltage and current spikes (within reason) along with Hitachi brand not surprised it still lives.
I liked Hitachi & Samsung drives but now we have no choice.
We are down to 3 manufacturers, if we ignore mergers/subsidiaries, WD, Seagate and Toshiba.
Not a great fan of WD or Seagate although I am sure the top end Enterprise Drives are good but expensive.
The domestic stuff does seem to be variable in quality and you have to navigate the various models to find the current 'best' drives.
I currently use Toshiba MD04ACA500 5TB drives which have , touch wood, been good and give no problems.

[guest]

I won't touch HGST drives with a bargepole. WD works for me, have had one failure in the best part of 15 years over maybe 20 disks - in the same timescale I've had multiple IBM/HGST drives die, several Seagates die (while lying about SMART status).

The ONLY benefit of IBM/HGST disks was a little-known one - if you used them with Thinkpads (or Lenovo business lappies when they bought IBM out) and password protected the boot sequence in the BIOS then the disks would be encrypted using AES256 (onboard hardware). None of that works now & you'd have to be clinically insane to use Lenovo after the Superfish (+ others) debacle.

If the SATA cable connector melted then the PSU is going to have future issues IMHO as the current taken to do that is going to be well out of spec. I'd be inclined to change it ASAP....

[sevenlayermuddle]

Interesting the one strand of agreement, neither of us like Seagate much.   

I've gone through a few seagates, including the 'server grade' original that shipped with the box, and all followed same sequence.  Smart status shows at least one or two bad sectors from new, then that gradually worsens , then there comes a point where it worsens at a frightening rate of knots, and I get scared and replace the drive.

Interesting suggestion about the strain on the PSU.   I've not yet powered the machine back on and won't do so until I have the time to do it methodically, checking all voltages.   ISTR it's not as easy as it might be to test a PSU, just to test open-circuit voltages you have to short two pins on the ATX connector.  Today the garden needs attention and all of next week's busy, so it'll be some while, but I will post update if/when I find out anything else of interest.

I may just replace the whole box.  It's normally left on 24/7 so I need to have confidence in it, which Rizla has just zapped, at least as regards the PSU (but I am grateful  ).      It's seen about 7 or 8 years service, during which time the energy cost of the circa 80-100W it consumes will have gone up more than I care to think about.  Maybe earned retirement, and replacement with something a little more modern.   

[Ronski]

The HP ProLiant Gen8 G1610T 4GB RAM MicroServer currently has £55 cash back offer on it.

[tonyappuk]

I recently built a new PC based on a Novatech barebones and had problems with a daisy chained SATA power cable. Novatech found it and replaced the cable (Shame! I should have found it.).  I assume these connectors are crimped and could it be a poor crimp or even a whisker bridging inside the plug. Just a thought.
Tony

[kitz]

OMG..  that is scary!

As I started reading I could imagine the smell as Ive had a blown PSU before, but as the thread progressed and looked at your image and realised how lucky you'd been that the damage wasnt more extensive.   I'd also not heard of this before, but having just had a look around it seems like its not an uncommon event with molex to sata adapters

There's quite a good thread here
http://www.stevenhale.co.uk/main/2012/05/sata-plug-melted/

There's some linkedpages, but note the mention of Amphenol and how there was a product recall in the US due to fire hazard. 
Also note how this seems to only affect the molex to sata adapters.

[loonylion]

I won't touch HGST drives with a bargepole. WD works for me, have had one failure in the best part of 15 years over maybe 20 disks - in the same timescale I've had multiple IBM/HGST drives die, several Seagates die (while lying about SMART status).

Large scale studies by backblaze.com show that HGST drives have the lowest failure rate of the major manufacturers when it comes to consumer drives. WD are close behind them, and seagates are awful. When confronted with the results, Seagate tried to blame the test methodology by saying that BB were using consumer drives in an enterprise environment, which is irrelevant because all the drives were consumer drives and they were all subject to the same conditions.

I wouldn't touch seagate if I was paid to at the moment; ever since they bought out maxtor (who were awful in their last years) they have been awful.

[sevenlayermuddle]

There's some linkedpages, but note the mention of Amphenol and how there was a product recall in the US due to fire hazard. 

Crikey, looks like I have been lucky indeed - could have been so much worse.  As Aardvark suggested earlier in this thread, combined with the fluke I backed it up to Blu Ray on Friday...  maybe it really would be a good idea to buy a lottery ticket this week, while it lasts.   

Knowing my more typical luck of course, the Blu Ray will turn out to be just a coaster when I try and read it.      It's got no media - music, TV stuff etc, but that can be recreated from original CDs.  It has all the critical stuff that really matters, SVN repos, email backups since the dawn of time, Bugzilla (bug tracking database) and the likes.   I think the SVN repo was actually on the suspect disk  so right now, I'm taking very good care of that Blu Ray.     

From that 'recall' link

The SATA side has the words Serial ATA on one side with the word Amphenol on the other.

That's a very general description, but it certainly meets the description of mine. And the vintage is about right too.  I definitely registered the system with HP because it was a spectacular bargain (just over £100) and with a spectacular 'server grade' warranty, with next-day on site service.  So they could have recalled if they'd tried.

I liked Tony's idea too, of a whisper bridge, which is kinda what's being discussed elsewhere.  I'm no expert on these things but the damage looks to me like it began at the cable-entry point of the connector, maybe a small outbreak of flame, then spread out up th cable, and towards thee edge of the connector, melting it as it went.