For the second time my array has gone offline and all disks were missing. (Including cache and the Unraid flash drive.) Also missing were the reboot and power down options.

 

The first time this happened, I couldn't get it to boot backup at all. It would get about halfway through Post and then just sit there. I was actually getting ready to upgrade the hardware so I went ahead and did that assuming the the likely culprit was the motherboard. (It's an old old motherboard and the audio chipset was already fried.)

 

The upgrade consisted of parts that had been in my main pc. An AMD Phenom II x4 925 processor, a Gigabyte GA-MA770T-UDP3, 16GB of ram (I forget brand, but decent), and I switched out the power supply with a 500w Enermax Liberty power supply that I had in another computer. (The old power supply was a basic Antec  500w power supply.)  Basically all different parts except for the hard drives.

 

The hard drives are 2 new HGST 6TB NAS, 2 WD 500GB greens, 1 WD 500GB black, and 1 Intel 180GB ssd for the cache. I should also mention that the WD drives are connected to the sata on the motherboard while the ssd and the 2 HGST drives are connected to an sata card in a pcie slot.

 

It's been running for about a week with no issues. Woke up this morning to find the array offline and all drives missing. Had to do a hard reboot and it did boot up just fine. All drives are back and the array started. A parity check is in progress from an unclean shutdown.

 

When it happened the first time I was in the middle of transferring a lot of files to the Unraid server. This time I think it happened during a scheduled parity check.

 

This tends to make me think that the problem happens under load, but the problem with that is that it doesn't happen every time. I have done a lot of file transfers and also a parity check within the last week and there were no problems.

 

I am new to Unraid and I haven't come across a similar problem in my searching. I really don't know what it could be.

 

Is a 500w power supply not enough? The Enermax is modular, do I need to add more power supply cables to split up the drives more? I could also swap it out with a 650w Enermax Liberty that I have in my main pc if I had to.

 

Any other ideas? I tried to get a diagnostics report before rebooting, but it came up as a 404 error. I'll attach the one after it came back up.

turtle-diagnostics-20151215-1036.zip

Definitely sounds like a power supply issue.  What is the exact model of the supply?

Definitely sounds like a power supply issue.  What is the exact model of the supply?

If it's this one, not single rail so it's really only giving half to the drives.

Definitely sounds like a power supply issue.  What is the exact model of the supply?

If it's this one, not single rail so it's really only giving half to the drives.

 

That looks like the one. I do not understand single rail.

Definitely sounds like a power supply issue.  What is the exact model of the supply?

If it's this one, not single rail so it's really only giving half to the drives.

 

That looks like the one. I do not understand single rail.

If you look at that page I linked, it is splitting the power available (12V1 and 12V2), giving half of it for video cards or whatever, and only half of it is available for your disks. Don't really see a 650 Enermax Liberty, but looks like the 620 on their site has the same problem.

 

See The Power Supply Thread

It looks like my power supply is actually an earlier version of the one you linked to. The power distribution seems to be just a little different. 384W(32A) for the 12v rails instead of 456W(38A). I probably bought it before 2009. The 650 is a little newer, but you are right, it is probably the same way.

 

I will look through the power supply thread later. (My morning has been eaten up by this problem.)

 

But what I would like to know is why doesn't the problem show more often? Like right now while it's running a parity check which is what I am assuming triggered it during the night.

 

Another question would be is if this is the problem, then what can I do about it for the time being since I can't get another power supply for awhile?

Power Supplies age, and lose capacity over time.  If its borderline then intermittent problems such as yours could result.

 

If you're handy with a soldering iron, you could cut the lines that are used for the video card power (which is a separate rail) and then splice them into another chain of sata / molex power connectors (You can use the 5v lines from anywhere -> their power draw is miniscule).  But, if you're not confident about soldering or identifying the wires, etc then don't do this because if you mess up you could destroy the hard drives and/or power supply

Soldering I can do. Identifying wires could be a problem, though. Once the parity check is done, I can take out the power supply and have a look.

Soldering I can do. Identifying wires could be a problem, though. Once the parity check is done, I can take out the power supply and have a look.

Meter to be perfect, but Yellow would be 12v, Red is 5v, Black is ground.  If you're going onto a chain of sata power cables, then you don't need to worry about Orange (3.3v) -> it isn't used. I take zero responsibility on any mistakes you make.  But it does work, and can allow you to continue to use a supply that you're not in a position to replace  https://lime-technology.com/forum/index.php?topic=23796.msg209556#msg209556

 

And, if you're already soldering together wires, then it would also be a good idea (although not strictly necessary) to permanently wire together any Molex splitters you might have.  At best, they make a terrible connection compared to a solder joint.

I'm still thinking about trying it, but I do have a question. You said the video card power is on a separate rail. So with 2 rails, what is rail 1 powering and what is rail 2 powering?

 

With the modular power supply, it would be easy to take 1 of the 12v yellow wires from the a video card line and splice into the 12v of a molex/sata line. (That is, if I am understanding what you mean correctly.)

 

However, I may just need to bite the bullet and get a new power supply. (Anyone want to buy a Ceton Infinitv4 and a Ceton Echo from me? lol) I took this one apart to blow the dust out and it looks like there's at least 3 popped caps and a couple of others that might be.

 

I also still need to check the other power supply I was using because I don't think it was a double rail (but I don't know that for sure). So that wouldn't explain why it had a problem. At the moment it's powering my HTPC and running fine.

I'm still thinking about trying it, but I do have a question. You said the video card power is on a separate rail. So with 2 rails, what is rail 1 powering and what is rail 2 powering?

 

With the modular power supply, it would be easy to take 1 of the 12v yellow wires from the a video card line and splice into the 12v of a molex/sata line. (That is, if I am understanding what you mean correctly.)

TBH, I would bite the bullet and get a new supply.  Had a look at the manual, and 12v1 is exclusively for the motherboard / cpu, and 12v2 is for the GPU and the drives, so splicing into the GPU cables won't make a difference.  Your other option would be to splice into the 12v connectors for the CPU, but since your CPU can draw ~8-9A you're not going to get much further ahead.

 

Not to mention the popped caps, which is definitely going to impact stability of the entire system

 

Yeah, I found that too. I'm not using the GPU power, so it wouldn't be a problem to take from there, but I'm not sure I want to take from the CPU.

Well, I ordered this.

http://www.amazon.com/gp/product/B00918MEZG

 

I'm considering cancelling and ordering the 450w instead. I'm not sure I need more than 37a. In the near future I don't see myself having more than about 8 drives. I will be switching to a 6th gen i5 and a new motherboard probably within the next year and I will probably be using a few pcie cards.

Power Supplies age, and lose capacity over time.  If its borderline then intermittent problems such as yours could result.

 

If you're handy with a soldering iron, you could cut the lines that are used for the video card power (which is a separate rail) and then splice them into another chain of sata / molex power connectors (You can use the 5v lines from anywhere -> their power draw is miniscule).  But, if you're not confident about soldering or identifying the wires, etc then don't do this because if you mess up you could destroy the hard drives and/or power supply

 

Not an expert here, but don't Power Supplies have some pretty dangerous capaciters in them?

Power Supplies age, and lose capacity over time.  If its borderline then intermittent problems such as yours could result.

 

If you're handy with a soldering iron, you could cut the lines that are used for the video card power (which is a separate rail) and then splice them into another chain of sata / molex power connectors (You can use the 5v lines from anywhere -> their power draw is miniscule).  But, if you're not confident about soldering or identifying the wires, etc then don't do this because if you mess up you could destroy the hard drives and/or power supply

 

Not an expert here, but don't Power Supplies have some pretty dangerous capaciters in them?

Of course...

Not an expert here, but don't Power Supplies have some pretty dangerous capaciters in them?

 

Define "dangerous".. 

It's not like he was told to take apart an old tube Tv (which could very likely kill him if he didn't discharge it prior to).

HV with minimal amperage, he may just get a prick or two.

 

More serious, and still likely the PSU, but any possibility his flash drive is becoming corrupt?

Since configuration is stored on the flash, unless this is only read at boot, and then store in RAM (uncertain).

Not an expert here, but don't Power Supplies have some pretty dangerous capaciters in them?

 

Define "dangerous".. 

It's not like he was told to take apart an old tube Tv (which could very likely kill him if he didn't discharge it prior to).

HV with minimal amperage, he may just get a prick or two.

 

More serious, and still likely the PSU, but any possibility his flash drive is becoming corrupt?

Since configuration is stored on the flash, unless this is only read at boot, and then store in RAM (uncertain).

If you actually read the posts, nowhere was it suggested to open the supply... Only to cut (splice into) other rails.

Not an expert here, but don't Power Supplies have some pretty dangerous capaciters in them?

 

Define "dangerous".. 

It's not like he was told to take apart an old tube Tv (which could very likely kill him if he didn't discharge it prior to).

HV with minimal amperage, he may just get a prick or two.

 

More serious, and still likely the PSU, but any possibility his flash drive is becoming corrupt?

Since configuration is stored on the flash, unless this is only read at boot, and then store in RAM (uncertain).

 

I only opened the power supply to blow it out. (It was very dusty even though I blow out my pc's every few months.) While I am not an expert with electronics, I am comfortable enough to do that. (I have fixed an LCD TV of mine by replacing popped caps... and without killing myself in the process! lol)

 

What Squid told me to do I didn't even have to have the wires connected to the power supply to do. (Because it's modular.)

 

As to the flash drive. I do not know if that could be the problem or not. I will say that it is a brand new flash drive that I bought specifically for Unraid. I have put it in a Windows pc and ran a quick check and it comes up as fine.

 

What I am wondering about is with the popped caps, how that would affect the power supply. I am getting the proper volts and everything turns on. My guess is that maybe I am not getting the amps that I should be? If so, would this then cause the power supply to use more electricity? I've only recently added a UPS, but I notice that the load on the UPS is around 153w when the Unraid server is doing nothing. The only other 2 things hooked up are an 8 port gigabit switch and an HDHR4 tuner.

If you actually read the posts, nowhere was it suggested to open the supply... Only to cut (splice into) other rails.

 

Admittedly you're are correct, point taken!

 

I only opened the power supply to blow it out. (It was very dusty even though I blow out my pc's every few months.) While I am not an expert with electronics, I am comfortable enough to do that. (I have fixed an LCD TV of mine by replacing popped caps... and without killing myself in the process! lol)

 

What Squid told me to do I didn't even have to have the wires connected to the power supply to do. (Because it's modular.)

 

As to the flash drive. I do not know if that could be the problem or not. I will say that it is a brand new flash drive that I bought specifically for Unraid. I have put it in a Windows pc and ran a quick check and it comes up as fine.

 

What I am wondering about is with the popped caps, how that would affect the power supply. I am getting the proper volts and everything turns on. My guess is that maybe I am not getting the amps that I should be? If so, would this then cause the power supply to use more electricity? I've only recently added a UPS, but I notice that the load on the UPS is around 153w when the Unraid server is doing nothing. The only other 2 things hooked up are an 8 port gigabit switch and an HDHR4 tuner.

 

Capacitors are used for multiples of things, and in a switching supply one of those is that they regulate and store voltage.

Basically it switches very fast pulsing a transistor, this creates a voltage, you store this and smooth it out with capacitors and regulate the output (this also has voltage regulators, resistors, coils, etc..) Some good information here https://en.wikipedia.org/wiki/Switched-mode_power_supply

 

I don't think your current draw would go up, no, however your voltage wouldn't be stable and certainly lead to issues.

 

New power supply is in. Because it's an intermittent problem and I can't purposefully recreate it, I'll only know if it's fixed after giving it some time.

 

If there are no more problems, I'll report back in a month. (If there are, I guess it will be sooner.) That also hits the monthly parity check since that's what the server was doing the last time the problem happened.

It happened again.

 

I tried to get the diagnostics, but it gives a 404 error. I'm attaching the syslog entries from overnight.

 

Looking in System Devices under SCSI devices, it does still show all drives. I also noticed that my share are still available. (I know they weren't the first time this happened. I don't think they were the second time, but my memory is not clear enough to be 100% sure.)

 

Edit: I may have spoke too soon. I am now having problems accessing the shares.

 

CPU Utilization is not showing a percentage. (The percent sign only, but no number, not even 0.) Memory usage looks normal.

 

It's Christmas day so i realize I may not get a response for awhile, but I will try to leave it like it is in case I need to check anything else while it's in this state.

Syslog.zip

I uninstalled the system stats plugin and my drives showed back up again briefly. I still couldn't connect to my shares with the message "A device attached to the system is not functioning".

 

All drives disappeared again and in the syslog and I am now seeing a slightly different message.

Dec 25 09:02:01 Turtle crond[1427]: exit status 127 from user root /usr/local/emhttp/plugins/dynamix.system.stats/scripts/sa1 1 1 &> /dev/null

I uninstalled the system stats plugin and my drives showed back up again briefly. I still couldn't connect to my shares with the message "A device attached to the system is not functioning".

 

All drives disappeared again and in the syslog and I am now seeing a slightly different message.

Dec 25 09:02:01 Turtle crond[1427]: exit status 127 from user root /usr/local/emhttp/plugins/dynamix.system.stats/scripts/sa1 1 1 &> /dev/null

post the full diagnostics

 

I uninstalled the system stats plugin and my drives showed back up again briefly. I still couldn't connect to my shares with the message "A device attached to the system is not functioning".

 

All drives disappeared again and in the syslog and I am now seeing a slightly different message.

Dec 25 09:02:01 Turtle crond[1427]: exit status 127 from user root /usr/local/emhttp/plugins/dynamix.system.stats/scripts/sa1 1 1 &> /dev/null

post the full diagnostics

 

I can't. It gives me a 404 error.

If you can still telnet in (or at console) in, type diagnostics.  Ignore any error messages.  Should be stored on the flash drive.

 

Barring that, post the entire syslog from /var/log/syslog