I am new to the NAS / Homelab community. Few months ago I managed to build and install small server running TrueNAS with 5x WD RED SSD SA500 2TB drives in RAIDZ1 configuration.
Few days ago the pool was marked as degraded with one of the drives (/dev/sde) marked as faulty with read/write errors. The drive was unable to run SMART test through TrueNAS interface, nor through CLI.
I checked joeschmuck’s flowcharts summary (BTW, thank you for these), and it lead to the conclusion the drive is faulty and it needs to be replaced.
Few hours later I did a scrub test and another drive (/dev/sdd) was marked as faulty while the previous one was back online without errors. Another Scrub test put offline different drive (/dev/sda), with the previous two showing no errors when running SMART tests (Short & Long).
Before the last Scrub, I put the latest faulty drive /dev/sda online and the result is a Heatlhy Pool.
When I was checking the temperatures of the drives marked as faulty I noticed weird spikes leading up to 62C for short few seconds.
Do you know what have caused this weird behaviour?
The NAS is used to run Plex and few lightweight containers to play around with self-hosting (Actual, InvoiceNinja, Vaultwarden).
All the SSD drives were bought at once from one seller. They are brand new.
The configuration of the NAS:
Board: Asrock N100M
RAM: 32GB Samsung DDR4
M.2 adapter is used to connect 5x SATA SDDs to the motherboard
While reading your post I kept thinking that I need to ask you if you are using any port replicators or weird cables since it reads like a cable/connector issue. Then I read:
So… if this is the only way you can run it, check the cables used. Check the power cables as well. If you are still having issues after that, try to change our your adapter. They should be cheap enough.
Any reason you are not using a HBA[1]? Your board seems to have available PCIe slots as well as two on board SATA ports. This would be the preferred and most reliable method.
These drive designations can change after a reboot. So if you rebooted while troubleshooting, it can very well be, that the drive with the errors was always the same.
Back up important data to a separate hard drive or other NAS.
Confirm that a high-quality power supply is being used and that the plug is securely connected.
Confirm that a high-quality SATA data cable is being used and that the plug is securely connected.
Confirm that the hard drive expansion card is properly cooled.
Confirm that the hard drives are properly cooled.
Run a SMART long test on all hard drives simultaneously and observe for any anomalies under full-speed read conditions, especially changes in the temperature of the expansion card and the hard drives themselves. The temperature of the expansion card may need to be measured with an infrared thermometer.
Use a hard drive performance testing tool to perform at least one hour of random and sequential read/write tests on the hard drives.
If your system passes all tests, you can treat the previous downgrade as an error report.
Hi ProfessionalAmateur, thank you for getting back.
No port replicators are used, and the cables are standard issued SATA cables that comes with motherboard. Will definitely check the connections once more, same with the power cables.
Honestly, HBA was not something I considered when I was building the server. Are there any recommendations which would be sufficient enough for my configuration? There are way too many options / cards.
Hi Farout, good point, I did restart the system once or twice between the scrubs, but the temperature swings are shown exactly on the three drives that had issues with Smart tests.
Would you say using HBA card (as suggested by ProfessionalAmateur) would be the preffered way to connect more drives to the system? These adapters / expansions cards are totally new to me.
In case you want some background reading on why HBAs are recommended over port multipliers.
As for a recommendation on an HBA; I use an LSI IBM sas9211-8i
Lot more options depending on how many drives you want. If you’re looking for a reputable seller, then https://www.artofserver.com/ is generally well regarded (no affiliation or sponsorship) & can be reached out to for a recommendation.
I prefere to first use the onboard SATA ports.
The onboard SATA controller usually doesnt need any active cooling.
If I then run out of ports, a SAS HBA in IT mode is the way to go + active cooling (important !)
These HBAs are plentiful on ebay and an 8 port HBA often doesnt cost more than one of these m.2 to sata adapters. but they need to be cooled (for example by a jury rigged 40mm noctua fan).
it needs to be a HBA
no RAID
in IT mode and latest firmware
An example would be the LSI SAS 9300-8i (you can connect SAS or SATA drives)
That m.2 to SATA adapter is a port replicator/multiplier. That’s often an issue. That’s why I mentioned and looked for it.I don’t get why it is recommended that often. It SHOULD work since m.2 is only PCIe[1] but it often doesn’t.
Recommending an HBA isn’t easy since it depends on the availability and your needs. Start with the two models mentioned here and go from there. If you have questions about a specific one, search the forum. I’m sure you will find a thread where the one you’re eyeing is mentioned and discussed. If not, you can always ask again. A lot here have had experiences with HBAs.
