My main Linux workstation started having serious errors pointing to memory issues but then even after I thought I had clearly rectified the memory issues both the boot disk and data disk started complaining about issues and the system was dragging. I built this system back in the end of 2020 to the beginning of 2021.
A few days ago I had updated the Fedora Linux OS with the procedure listed at this URL https://docs.fedoraproject.org/en-US/quick-docs/upgrading-fedora-offline/ to go from Fedora 41 to Fedora 42 successfully. I have used this procedure many times going from Fedora 37 onward on my workstation and some VMs so I don’t think that was an issue.
Since so many components were complaining I checked the power supply. I didn’t feel like tearing down one of my legacy workstations and taking the power supply out of that for a test as I wanted to keep them whole to sell or pass on to a deserving soul, so I purchased a new 1050 Watt power supply.
The IPMI logs had some generic power supply warnings that I think were more related to not turning on an item in the bios than actual hard errors. Replacing the power supply I did notice that there was an issue with the 8 pin cable going to my XFX AMD Radeon™ RX 580 graphics card with one pin clearly out of the socket but that wouldn’t have caused the issues I was seeing.
The disk errors persisted when I replaced the power supply and though the IPMI log power supply general error persisted a couple of sensor readings did go from red to green and this is in the VCCM area that controls the voltage to the memory. I noticed this sensor go red when I had the faulty memory in there as well, which I shall remember from here on out and I never considered as a troubleshooting step or indicator. Pays to look up those settings if you don’t know them.
I blew out all of my RAM slots PCIE slots, fans etc. with canned air, I even had to alcohol wipe down a stick of RAM because I had two indicating they had issues. One stick of RAM came back good after that but the other remained failed.
Now feeling I had successfully concluded the bad RAM issue. I tackled the hard drive problems. Switching and reseating the slots cables etc. did take away the errors on my data drive which is of course my “/home” but not the errors on my boot drive. Switching back to the original PCIE slot the same issue persisted with my boot drive.
I didn’t see any fresh errors from the SMART disk utility and the SCSI sense errors seemed to clear but it appears that there’s still some block errors fsck didn’t fix. I could boot into live Workstation Fedora 42 media mount and perform an fsck of /root to fix those block errors but I haven’t been feeling well and the reboots take forever since I have my SAS HBA controller that takes while to boot as it interacts and makes itself available to the motherboards bios and I have to constantly go into the bios or hit F11, flip over the PC etc.
Since this is the weekend I can’t order a SAS 15k 12gps HDD to match my existing boot drive so I picked up a SATA SSD for my boot drive temporarily. The way I build my personal systems the last 30 years is separate boot and data drives and with Linux it makes it simple to upgrade the OS.
I just need to install whatever version of Linux I have, mount that to the boot disk I have /root, /boot and swap (optional these days) make sure I use the same user name etc., update fstab and I’m up and running with all my configs once I login with that user name. I also copy that data to my NAS systems so in case of emergency I can restore from there. Then I make sure I install the same packages and if I need to modify and files in /etc I have copies of them most of the time somewhere on my data drive.
The other issue I had is that my other workstation a Windows 10 system kept having issues burning the Linux ISOs/DVDs with either Rufus or Balena Etcher utilities. I even tried the DD option with Rufus which is at the block level but that didn’t work either.
I wanted to burn a version of the Fedora 42 server onto USB instead of the workstatiion flavor of Fedora because the workstation installer didn’t let you select the filsesystem type, they just defaulted to btrfs filesytem and I wanted to use EXT4 and LVM. I like the idea behind btrfs and will play with it later.
I ended up installing Workstation to the SSD drive booted from there and installed Balena Etcher and then created a USB Fedora 42 installation boot USB. Then reinstalled with the server version.
After installing Fedora 42 Server version to the SSD and mounting my data /home drive I installed my usual SW packages, made sure desktops and desktop switchers worked and I was back in business. Took a little too much time as I had to go to the store to get my SSD and power supply plus each boot was like 10 plus minutes, reconfiguring the power cables, plus I had to install some new RGB case fans on the bottom . This is where my penchant for redundancy comes in as I had duplicate AsiaHorse Matrix 360 fans to swap in.
Next I’ll mount and fsck my 15K SAS boot drive at some point when I have time. Could be a while since I have a lot of projects and fires in the works right now! Then I’ll determine if I need to order another drive. Sometime this year I had intended to up my NAS capacity and upgrade the NAS utility SW as well. I get more issues with my workstations that may be a project as well but with inflation and tariffs I will need to be diligent and keep and eye on prices and do my research on motherboards, CPUS and drives and possibly HBAs.
Linux workstation specs just in case you’re interested:
- Case:
- LIAN LI O11 Dynamic XL ROG case
- Pwer Supply:
- Thermaltake 1050 Watt Toughpower GF A3 power supply
- Graphics:
- AMD Radeon RX580 8gb
- Motherboard:
- ASROCK x299/WS IPMI motherboard
- Memory
- Corsair VENGEANCE® LPX 64GB (4 x 16GB) DDR4 DRAM 3000MHz C16 Memory Kit
- PCIE HBA disk controller:
- SAS 9341-8i 12 Gbps PCIe SAS/SATA RAID Adapter
- Disks:
- HGST 2TB SAS 12Gb/s 3.5
- PNY 1tb SSD
- Seagate Exos 7E8 8TB 7.2K
- PCIE Network controller:
- INTEL 10GB DUAL PORT PCI-E NIC NETWORK ADAPTER
- CPU Cooler:
- iCUE H150i RGB PRO XT Liquid CPU Cooler
- Case Fans:
- Asiahorse Matrix 350 fans
- OS:
- Fedora 42
