Katherine Posted November 17, 2016 Share Posted November 17, 2016 I was transferring data to the array and the transfer froze for a second. After the transfer completed, I checked unRAID and it was reporting a red X on one drive. Checking the drive the following receives a warning: 199-UDMA_CRC_Error_Count: 2 I ran an extended SMART test on the drive along with a round of pre-clearing. The drive passed SMART test and successfully completed the pre-clear. The question is should I send the drive back to manufacturer (HGST)? Can I even create an RMA based on UDMA CRC Error Count? More importantly, is it even safe to put the drive back into the array? Any advice would be appreciated. Quote Link to comment
RobJ Posted November 17, 2016 Share Posted November 17, 2016 The drive is fine! That usually indicates an issue with the SATA cable or connectors, so I would replace the cable with a good one, and check ALL cable connections (both ends of the cables) for tight connections. It's also possible there was a power glitch. CRC errors just indicate that a corrupted packet was detected, and they're always re-sent. A CRC error does not usually cause a disabled drive. I suspect something else happened, and the evidence may be in the syslog that covered that moment. Maybe there was a BIG power glitch! 2 2 Quote Link to comment
Katherine Posted November 17, 2016 Author Share Posted November 17, 2016 Thank you, Rob. I made my own SATA power cables from el cheapo ebay stuff and I'm using 430W PSU for 8-10 drives so its entirely possible its one or the other. I guess its time I replaced the PSU. Quote Link to comment
S80_UK Posted November 18, 2016 Share Posted November 18, 2016 Check the cables first. For a basic server 430W is normally fine with more drives than you have provided it can handle the peak startup currents for all drives at once. I have had a few glitches over the years - it has always been power cables or SATA cables that were responsible. 1 Quote Link to comment
Katherine Posted November 18, 2016 Author Share Posted November 18, 2016 Cables it is then! I'm thinking of buying this to split power: http://www.newegg.com/Product/Product.aspx?Item=N82E16812400127 Previously, I was using SATA and molex adapters to power the drives but considering my Cosair 430W PSU is single rails design, it shouldn't matter if I power all the drives by splitting power from SATA adapters? Am I correct in assuming this? Or should I continue using both SATA and molex adapters to balance power distribution? Quote Link to comment
John_M Posted November 18, 2016 Share Posted November 18, 2016 Since your power supply has a single 12 volt rail there's no balancing to be done. All the yellow 12 volt wires meet at the same point inside the supply. The only (minor) consideration is to make sure you don't draw more current down that single 12 volt wire than it's rated for. With what you're proposing to do it won't happen but you don't want to be connecting splitters to splitters to splitters, for example. Quote Link to comment
Squid Posted November 18, 2016 Share Posted November 18, 2016 Previously, I was using SATA and molex adapters to power the drives but considering my Cosair 430W PSU is single rails design, it shouldn't matter if I power all the drives by splitting power from SATA adapters? Am I correct in assuming this? Or should I continue using both SATA and molex adapters to balance power distribution? Presumably, you've got separate lines for the molex connectors and the sata connectors coming off of the drive. If you have to split the power to handle the number of drives, even on a single rail supply, you really want to balance the load between those sets of wires from the supply. You don't explicitly specify the model # of the supply, but if its a CX430, then you've got 32A on the 12v line available when the supply was new. Round figures, you can count on a drive taking up ~2A during spinup's, so if/when all the drives spin up together (parity checks / rebuilds, etc), then you're drawing 20A. But you also have to remember that the CPU also draws from the 12V rail, and depending upon the CPU it can pretty much be 8-10A itself when not idle. So you worst case scenario is a draw of around 28-30A on a supply that can handle 32A when new (and as supplies age they cannot handle their rated draw). Entirely up to you, but IMO the supply is a bit small to handle 10 drives. Quote Link to comment
Katherine Posted November 18, 2016 Author Share Posted November 18, 2016 Thank you for breaking it down, Squid. I agree with you. I'm pushing my luck with my current setup: CPU: Intel Xeon X3460 (90W) M/B: Supermicro X8SIL RAM: 16GB (4 x 4GB) HDD: 8 x HGST 4TB (7200RPM) RAID: AOC-SASLP-MV8 To boot, the case (Rosewill RSV-L4500) has 8 fans (6 x 120mm, 2 x 80mm) daisy chained with molex cables. I'm going to upgrade to a newer PSU before I add the remaining 2 WD Green drives to the array. Quote Link to comment
bombz Posted July 8, 2018 Share Posted July 8, 2018 (edited) WDC_EARS UDMA CRC error_output - RAW VALUE = 8731Power on hours = 7y 9m 4d 16h This is on my LSI card, but I changed the SATA cable to my on-board sata port (to swap the cable) rather then using the LSI breakout cable and still prone to the error. I think due to the age of the drive it needs to be replaced to be honest. Let me know your thoughts Thanks Edited July 8, 2018 by bombz 1 Quote Link to comment
guruleenyc Posted February 8, 2019 Share Posted February 8, 2019 Yeah I am seeing this on a couple drives as well, especially when in high utilization. I have changed the SATA breakout cables as well and it did not resolve it. Perhaps the drives are going... 1 Quote Link to comment
feralcoder Posted September 4, 2022 Share Posted September 4, 2022 This is a bus problem, not a storage problem. It COULD be related to bad drive HW, but it also could be bad signal or power connection, or insufficient / bad power. The least likely of those problems, I believe, is the drive itself. Also, compared to the slimmer chance that it's a drive's onboard controller, it's at least as likely to be RAID or drive controller or expander cards. That said, people seem to want to believe it's the drive since the count appears in a drive's smart output. In my experience it's far more likely to be flaky or insufficient power, or drive controller / expander. For example, at-home NAS setups frequently don't adequately cool the controller and expander cards. Until I put large fans literally ON EVERY controller and expander, I had flaky data problems all over the place. Also, it's a good idea to have an extra power supply on hand for debugging possible power issues. It doesn't have to be big - you can power a drive by itself, on its own power supply, for testing. And especially for ppl experiencing these same "UDMA CRC Error Count" problems on multiple drives, it's almost certainly not your drives. 2 1 Quote Link to comment
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.