Poor quality SilenX fans

[BingBangBop]

Today is not a good day!

About a month ago I switched out all my120mm fans for SilenX on my server: 2 for the processor, 1 case, and 2 for HD's. About a week ago, I noticed that the case and one processor fan had failed. No big deal, there is redundancy. Today, I found that the machine had shut down. I powered it up and the other 3 SilenX fans had all failed. So I replaced them with the old fans and I thought everything was good since the machine would still power up.

Well, My HD's obviously over-heated because I now have multiple HD's that continuously click every 10 seconds or so. They are seen by the bios, I can get data off them but it is all corrupted. This is what I call a bad day!

 

[ddrueding]

That does indeed sound like a bad day. SilenX never had the best rep IMHO, and this drops it considerably.

 

[Stereodude]

That sucks. I don't think I have any SilenX fans in any of my systems.

 

[LunarMist]

Were the SilenX fans qualified for that server chassis? A company with such a kooky name does not inspire confidence. :cat:

 

[MaxBurn]

I read somewhere that these fans have a pretty high rate of failure. They were saying it was limited to the 120mm models, which I agree with as I have four and two locked up on me in my p180. I had six more 92mm ones in my server and they never had a problem. What it was is there is a lack of lube on the bushing, you can feel by turning it by hand there is a lot of resistance. You can pull the fan apart and put a drop of motor oil on and it will run again fine. You actually grab the fan blades and pull them out of the fan housing.

 

[BingBangBop]

No, the SilenX fans were not server-certified. The brand is well known for moving lots of air relatively quietly. Really, I didn't even consider that they would all fail simultaneously without the normal screeching warning. Redundancy is almost always better than long MTTF's and I had redundancy.

If anyone thinks I'm about to try to repair these fans and put them back in service, are totally server-certifiable.

I've spent the day trying to recover the data. Still not done. One of the dead drives is my main backup. I've got an old backup drive that was removed and stored (and I'm using that data right now so I'm not totally down for a long time period), but that is old data and I much prefer keeping the data as new as possible which is why I am at least trying data recovery.

 

[time]

Were the SilenX fans qualified for that server chassis?

What sort of bullshit question is that?

A company with such a kooky name does not inspire confidence. :cat:

You're right. You have a kooky name and you don't inspire confidence either.

 

[LunarMist]

Were the SilenX fans qualified for that server chassis?

What sort of bullshit question is that?

I don't mean to cause trouble, but much of my recent life is about validation and qualification. I was training someone to conduct a risk assessment just Friday evening. First questions when a failure occurs are was the vendor qualified, what were the functional requirements, was design verification completed successfully, and so forth...

 

[LunarMist]

You're right. You have a kooky name and you don't inspire confidence either.

The username is a joke from a vendor audit I conducted some years ago. There is too much PII here as it, so I won't go into the details .

 

[time]

It was still an inappropriate question. Do you really believe that is how you should manage risk, by blind process rather than proper analysis?

The issue with the fans had absolutely nothing to do with the case, yet you were prepared to create a faux association between them; good for finding legal liability faults but quite irrelevant to ensuring the success of a project.

In any case, the only way a PM could 'qualify' the fans would be to consult the manufacturer's claims. That's a very basic first step that should be left to engineers. Testing the veracity of the claims is something that seems to be no longer attempted.

Sorry, it's one of my current pet areas of interest: why large projects usually fail, despite or perhaps because of modern methodologies.

 

[MaxBurn]

Really, I didn't even consider that they would all fail simultaneously without the normal screeching warning.

That is a pretty bad coincidence. Mine didn't make a screeching noise either and I wouldn't assume it would after looking at these bearings. They did make a moaning noise though, really low pitch. As far as my fix goes I didn't see any wear, just lack of grease/oil on the shaft so I fixed that. That had to be around November last year. Not like these things had bearings in the first place, just a plastic bushing on a metal shaft. My fans are all redundant too but I am also using bios fan monitoring to alert me.

 

[Chewy509]

Sorry, it's one of my current pet areas of interest: why large projects usually fail, despite or perhaps because of modern methodologies.

I've been doing some casual reading on this topic as well. From what I can gather it boils down to:
1. Lack of communication between stackholders (the client and vendor).
2. Management (on both sides) being dollar driven rather than solution/outcome driven.
3. No plan F. (The "we're f**ked plan, how do we get out alive"? plan).
4. Lack of acknowledgement that some things can not have it's time budget known during initial planning phases.
5. Lack of understanding of the domain which the solution will operate in.

PS. My interest has more been from the software development lifecycle perspective and software development projects, and have mainly been reading IEEE published literature.

 

[time]

1. Lack of communication between stackholders (the client and vendor).

The larger the project, the less clearly the client and vendor are delineated. I wonder if that also is part of the problem? What I'm trying to say is, the client drives the solution to the extent of embedding innumerable consultants, analysts and PMs, and the vendor responds in kind (as a sort of biological defense mechanism?). There's plenty of communication, but anything useful is lost in the noise.

2. Management (on both sides) being dollar driven rather than solution/outcome driven.

That's the inevitable consequence of modern project management principles. I'm still amazed that engineers are supposed to provide estimates at a subatomic level so that astonishingly ignorant people can pick and choose, thereby corrupting the integrity of the design. If you built bridges this way they'd fall down. Actually, you can substitute the word "software" there. If a project needs to be re-engineered to cut costs, it's a job for engineers.

3. No plan F. (The "we're f**ked plan, how do we get out alive"? plan).

I've hardly ever seen even a plan B. Gantt charts don't show anything other than 'The Plan'. Not sure what the point in risk assessment is when you assume you've resolved all the risks and fail to develop a backup plan, but that seems to be the norm now.

4. Lack of acknowledgement that some things can not have it's time budget known during initial planning phases.

If you push that line, you'll just be replaced. I've been shocked on the rare occasion when a manager admits something like that, because I'm usually hearing from some f***wit how accurate IBM is with their estimates, or some such bollocks.

I wonder how many people realize how frighteningly realistic Dilbert is? The unrealistic bits are where people get away with telling the truth to the boss' face.

5. Lack of understanding of the domain which the solution will operate in.

You mean, "don't have a clue what they're developing or what it's for"? See, this goes to the heart of my unease. There's never any shortage of business analysts these days, because apparently modern developers are no longer capable of intelligent thought (maybe they're right?). Everything is documented to death, complete with random UML and other whimsies, and yet, the subject domain seems to remain a mystery to all involved.

Maybe, just maybe, the methodologies are simply crap? Maybe everyone should go back and read "The Mythical Man Month"? I certainly don't have any faith in the IEEE.

 

[Chewy509]

Maybe, just maybe, the methodologies are simply crap? Maybe everyone should go back and read "The Mythical Man Month"?

I'm starting to think that as well...