RS_APT

To confirm and address your question regarding the hardware: I was mistaken in my original post. This is indeed a single-socket ASUS X99-DELUXE motherboard running a single Intel Xeon E5-2697 v4 CPU, exactly as reported in the diagnostics! To clear up the backstory: the initial crashes that started this whole issue happened while actively using the original 10Gb NIC under load. Per your advice, I left the 10Gb card removed and waited for the array drive rebuild to complete 100% cleanly over the 1Gb Intel interface before touching anything else. Once the rebuild finished, I enabled remote syslogging and began working through the hardware. Here is the exact step-by-step timeline of what occurred: Hardware Changes & Boot Failures: Performed a hard shutdown, installed a replacement Broadcom 10Gb NIC (same model, different card), and re-installed the HBA. Powered back on, but the system failed to boot. Rebooted again—still not working. Isolating Cards & Storage Glitches: Removed the HBA and rebooted—still didn't come up. Moved the NIC up a PCIe slot and rebooted—system finally came up, but 1 drive was disabled. Shut down and reseated the SATA cables. Powered back on—now 2 drives were down. Cable Replacement & HBA Re-install: Removed the NIC completely and replaced all SATA cables connected to those two drives. Rebooted—all drives came back online cleanly. Turned off, re-installed the HBA, and restarted—system remained stable and online. Re-testing 10Gb NIC & NFS Failure: Re-inserted the replacement Broadcom 10Gb NIC into the second slot from the bottom and booted. Motherboard displayed POST code FF (OS loaded), pulled a DHCP IP, and SMB shares worked fine. Issue: NFS shares completely broke. Restarting/toggling NFS did not help, and syslog showed errors (bzmodules). USB Drive Corruption & Repair: Removed the 10Gb card and rebooted. Realized the repeated hard shutdowns had corrupted the USB flash drive for the 3rd time during this process. Re-flashed/repaired the USB drive using the Unraid Flash Creator tool and restored my /config backup. Current Status: System booted up clean—syslog confirms zero decompression errors, drives are healthy, and SMB/NFS services are fully restored. The 10Gb NIC is currently removed/not configured for traffic yet, as I wanted to confirm baseline stability first before risking being back at square one. Since the initial instability started under heavy 10Gb network load, I am cautious if i even want to attempt it again hahaha. Its been hours and hours of troubleshooting. I have an export of my syslog file attached. How would you recommend re introducing 10gb again without blowing my NAS up lol ? unraid.log

pls see attached, I have had a look through these but cant see anything obvious. However due to need of reboot or .bak the network conf file to get it to boot i dont think there is much in the logs. As i have the 10gb NIC removed to restore system functionality. tower-diagnostics-20260907-1911.zip

Hey everyone, Hoping someone here can sanity-check an issue I’ve been wrestling with for a while. My Unraid server keeps crashing, especially during heavy drive activity, and I'm pretty sure it traces back to my 10Gb network card. My setup CPU: Dual Xeon E5-2697 v4 Motherboard: Dual socket board (enterprise) Primary NIC: Onboard Intel 1Gb (eth0 for management) 10Gb NIC: Broadcom BCM57810S dual-port SFP+ (bnx2x driver) connecting point-to-point to my Plex box Slot Placement: The 10Gb card is in the bottom PCIe slot, which routes through the motherboard chipset. How it started vs how its going At first: WebGUI died, but Ping/SSH stayed up Initially, whenever there was a bit of heavy network traffic over the 10G link, the WebGUI would just freeze up completely. The weird part was that I could still ping the server and SSH into it just fine. Checking lspci -vvv in the terminal gave me this error over and over: pcilib: sysfs_read_vpd: read failed: No such device I added options bnx2x disable_tpa=1 disable_fw_vpd=1 to my /boot/config/modprobe.d/bnx2x.conf file, which seemed to stop the random WebGUI freezes while the server was idle. Now: Complete system freezes & boot issues Things have gotten way worse recently: Starting an Array Rebuild instantly kills the server. As soon as I start rebuilding a drive, the whole box hard freezes within seconds. No WebGUI, no SSH, and no ping response at all. Complete dead lock. Reboots are super unreliable. Halting or rebooting from the GUI often leaves the server completely powered off or stuck in limbo instead of spinning back up. Sometimes on boot it takes forever to initialize or shuffles my interface network assignments (eth0 through eth3). what actually is working right now: If I physically unplug the 10Gb DAC cable and remove from PCIe lane, let the drive rebuild run purely over the onboard 1Gb Intel port (eth0), the server is 100% stable and the rebuild completes without a single hiccup. As soon as the 10G cable is handling traffic during heavy disk writes, the system dies. My questions Has anyone else run into total system deadlocks using Broadcom 57810S cards in chipset-shared PCIe slots? Is there any driver or kernel parameter that can keep this card from losing its mind when PCIe bandwidth gets congested? Since my CPU-direct PCIe slots are blocked by my GPU/HBA, am I better off just binning this Broadcom card and picking up an Intel X520 or Mellanox ConnectX-3?