November 11, 20178 yr I just got done replacing my parity drive with a larger one and turning the old parity drive into a data drive. The old parity drive is still pretty new and never showed any problem. I started writing files to it as a data drive, and it's been X-ed out with the syslog showing a bunch of errors. The short SMART report shows nothing wrong. I didn't so much as touch the cabling to that drive, I just added a new parity drive. I checked the cabling and didn't see any problems, then just as a check I precleared the old parity drive and it passed without errors. Then I placed the drive back into service and let it rebuild. After it rebuilt (uneventfully), I wrote tens of gigs onto it without errors. Then later, after everything had spun down, I tried some more writes and got errors almost immediately. The only other things notable is I just recently upgraded to unraid 6, and when I formatted the ex-parity drive I chose XFS for the first time. Logs are attached. Can anyone see a clue as to what's going on? Thanks! tower-diagnostics-20171111-0916.zip Edited November 18, 20178 yr by jhyler
November 11, 20178 yr Looks like the drive (16) is dropping offline. Could possibly be a poor / loose cable. More likely though its the Marvel controller. LSI based controllers are the recommended ones for v6. It *may* be possible to eliminate this problem by moving if at all possible the 2TB drives and the 10TB drive off of the Marvels and put them instead onto the motherboard. But, Marvel has terrible compatibility with drives and 64Bit OS's
November 11, 20178 yr Author Thanks for the quick reply! This is in a Norco case which supports 20 drives of which 17 slots are full of disks, so if the controllers (two Supermicro SASLP-MV-8s) are suspect, my thought is to take the system down until I can replace the controllers. Interesting that preclear doesn't show the issue, maybe because its patterns of reads and writes is quite different from real world operation. Thanks again! Edited November 11, 20178 yr by jhyler
November 11, 20178 yr Author Wow! I thought I'd just check the boards to learn what controller card(s) to get, but it turns out those threads get real complex very fast or are marked "old, don't trust". Can anyone give me some pointers on good SAS/Sata controller cards that work with Unraid 6 that (1) Support at least 8 drives each, (2) Available low-profile bracket, (3) Support large (i.e., 10TB) drives, and (4) Work with Unraid out-of-the-box (i.e., no BIOS flashing required) Thanks! Edited November 11, 20178 yr by jhyler
November 11, 20178 yr ...or the newer Adaptec HBA1000-8i - works out of the box (info from bonienl) Edited November 11, 20178 yr by Zonediver
November 11, 20178 yr Author Thanks, all. I'm thinking I will go with the 9207-8i. My only concern is that my motherboard only has two PCIe slots, one x16 and the other one being x4 on x16. These being x8 cards, I'm expecting a bottleneck on the one - but it shouldn't be worse than what I already had, right?. Oh well, if it gets me past this emergency maybe Santa will bring me a new motherboard. I appreciate all the help. You guys are the best. Edited November 11, 20178 yr by jhyler
November 11, 20178 yr 18 minutes ago, jhyler said: Thanks, all. I'm thinking I will go with the 9207-8i. My only concern is that my motherboard only has two PCIe slots, one x16 and the other one being x4 on x16. These being x8 cards, I'm expecting a bottleneck on the one - but it shouldn't be worse than what I already had, right?. Oh well, if it gets me past this emergency maybe Santa will bring me a new motherboard. I appreciate all the help. You guys are the best. Then give us a Like Edited November 11, 20178 yr by Zonediver
November 18, 20178 yr Author Update - Replacing the controllers did indeed correct the problem. Thanks, all.
Archived
This topic is now archived and is closed to further replies.