kaydaryl

Thank you to everyone for the input! In addition to updating the BIOS on the 9300-16i (which is installed as we speak) I also updated my motherboard's BIOS to the latest version, and then re-ran the full stack of memtest86+ for a few more loops with no errors. I can now say I'm over 2 days of uptime and have far fewer headaches with the HBA-connected cards, including much faster r/w speeds than I originally did. To stress it I've since been pushing writes to the array with junk data to see if I could recreate some of the unpredictable freezes of the system and haven't had any problems thus far. While >2 days (so far) isn't definitive to call it super-stable yet I wanted to update all of you to say thank for giving me the right rabbit holes to explore. I suspect it was oddities in the 6.12.0-.1 release immediately followed by installing an HBA with an outdated BIOS that was the root problem.

The motherboard came as part of a Microcenter CPU+mobo bundle and was purchased planning for an HBA to offload most connections anyway. After reassessing the HBA I found that while the firmware was current the BIOS was quite a few revisions old. I've updated the BIOS on both controllers placed it back into the server to see if that helps stability, else I'll remove it and run the server with a single HDD connected via SATA for a week to assess. Not sure why the memtest run was so short, I expected something 12+ hours with everything enabled in the memtest ISO. I assume it was only 1 loop but I didn't see anywhere to increase the number of iterations. I'm really hoping it was a perfect storm of new Unraid user, issues with 6.12.0, and an HBA with outdated firmware/BIOS!

- HBA is needed because my Define 7 has plenty of room and I'd like to fill that. - My mobo only has 4 onboard SATA ports. For reference my NAS has 6 slots already filled. - The 6x2TB in my current array are because I had them lying around; I am not planning to use them long-term. While slow, they have been helping me understand how Unraid manages files without any harm to important information. - My current NAS is maxed out on slots (3x10TB and 3x14TB) and is full of content. With Unraid acting as the server it doesn't make sense to get larger drives for the NAS. The plan is that once I have the confidence to add 3 20-22TB drives to the server I can then migrate data off of the NAS drives (no parity) and then add those drives as well, adding a parity drive once I have the capacity to maneuver that. With a single parity drive that puts me over 4 HDD, necessitating an HBA.

I've caught another reboot seen in the attached log at Aug 4 18:40:43. This time the syslog reports issues with the HBA before an unexpected reboot occurs as was seen here. After the reboot no HDD are seen as mounted. I've removed the HBA from the system and connected a single HDD via SATA to the mobo to see if it's more stable (no important data are on these HDD yet, don't care if I lose anything). Aug 4 18:40:43 Server kernel: mpt3sas_cm0: SAS host is non-operational !!!! Aug 4 18:40:43 Server kernel: mpt3sas_cm1: SAS host is non-operational !!!! The HBA is running 16.00.12.00 firmware, not sure if there's newer firmware for that model, will report back if I get further errors with the HBA removed! syslog

Hi Jorge, I've re-run memtest86 with the latest build from their site and passed with both sticks in place, would it be beneficial to run it again with only 1 stick in at a time?

Hi Jorge, I had run memtest with an AEMP profile enabled when I initially built the rig and passed with flying colors, would it be worth it to again?

Hi, I have a new server that I'm hoping can eventually replace the Ubuntu server and NAS setup that I have used for years but am still struggling with stability issues that are causing the server to need to be physically rebooted too often, sometimes it's not even online for an entire day! I'm still keeping important files on the NAS and pointing Dockers to that mount, but want to be confident with my equipment and see stable uptime before committing to migrating everything over for good. I can see that a crash occurred again this morning before 4:29AM however there doesn't seem to be any indication of what caused it, unfortunately. This morning I saw drives were missing from the array so I powered down (8:16AM in the syslog), reattached the HBA's PCIe, and took a fresh download of server-diagnostics. For what it's worth, this occurred before I even added the HBA card. I chalked it up to the complaints here and on Reddit about 6.12 being unstable, but am still experiencing problems on 6.12.3. Specs: Mobo: Asus Z690 + 12600K RAM: 2x16GB, AEMP disabled (stock 4800 for DDR5) HBA: 9300-16i (PCIe power attached) with 80mm Noctua + 3D-printed shroud (I find SAS host is non-operational and drives disappear if PCIe isn't attached) PSU: Corsair RMx (2021 edition) 650W Cache: Intel 670p 2TB HDD: 5 2TB HDD + 2TB Parity Syslinux.cfg details: append intel_iommu=on iommu=pt acpi_enforce_resources=lax pci=noaer pcie_aspm=off pcie_acs_override=downstream,multifunction isolcpus=0 initrd=/bzrootq Please help! syslog.log server-diagnostics-20230804-0841.zip