I'm having some frustrating issues with my HP MicroServer Gen8, which is currently running the latest version of Proxmox VE. I've been experiencing recurring SATA errors where my drives randomly switch to emergency read-only mode once or twice a day. It usually starts with one drive failing, and then other drives seem to follow suit shortly after. Here are the error messages I'm encountering:
- `ata1.00: failed command: WRITE DMA`
- `ata1.00: error: { ABRT }`
- `sd 0:0:0:0: [sda] Sense Key: Illegal Request`
- `Add. Sense: Unaligned write command`
- `I/O error, dev sda, sector 2048`
- `EXT4-fs: I/O error while writing superblock`
In terms of setup, I run my system off a SSD connected through the ODD port, with GRUB on a USB drive. The front bays have a mix of drives: 2 x 4TB WD Red, 1 x 4TB Seagate, and 1 x 12TB Seagate, with backups on USB drives. All drives are formatted with Ext4, mounted by UUID, and used within Docker containers on a single Ubuntu container.
I've already checked the SMART values multiple times, and they look good with no reallocated or pending sectors. I've also verified all cables are secure, cleaned connectors, and disabled the WD idle timer just in case. Here's some additional context:
- I upgraded to an Intel Xeon E3-1265L v2 CPU.
- The server has 16GB of non-HP RAM.
- I've jury-rigged a SATA power adapter for the ODD bay, but the system SSD is functioning without issues.
The BIOS is set to AHCI mode, with SATA power settings on maximum performance, and both BIOS and iLO are up to date.
I'm hoping to hear if anyone has had similar experiences or can offer some solutions. Thanks in advance for any help!
3 Answers
Remember that the Gen8 only has two of its four bays supporting 6Gb/s speeds, while the others are limited to 3Gb/s and even 1.5Gb/s for the internal blue port. Mixing drives can definitely lead to problems, especially since the B120i isn't known for stellar performance even in what they call pseud-HBA mode. Moving your system drive to one of the faster bays might help stabilize things for you.
It sounds like your B120i controller might be causing those issues. Even when set to AHCI mode, it doesn't truly pass through drives properly. I've noticed that mixing different drive brands can lead to unaligned write problems like the ones you're encountering. I recommend booting from a USB and monitoring the system logs with `dmesg -w` while you stress the drives. If the same issue pops up quickly, it's likely a firmware problem, not the drives themselves. Sometimes, switching to a proper HBA, like a P222 or LSI 9211 in IT mode, can resolve these issues overnight.
I ran those tests, and you're right. The logs showed similar errors, but they appeared much more frequently. I actually got the MicroServer just a couple of weeks ago, and it was running fine until recently.
Try swapping the drives around to see if the issue is due to a specific bay or the drives themselves. It's possible that after all these years, your box is starting to give out. With the upgrades, it's definitely pushing its limits—especially with that Xeon CPU. I have three Gen8s, and while they've been solid, I've noticed two of them are starting to act up, so I may need to replace one and salvage parts from it.
I did try that a couple of weeks ago, but the SATA3 connections were the ones failing most often, even with different drives. I suspected power issues since I read that the Gen8 backplane can struggle with higher power consumption. I didn't think of rolling back to the old CPU to see if it changed anything—thanks for the tip!
So, could it be tied to the different SATA speeds, especially since my system drive isn't in one of the prime bays? I'd be okay with relocating it to bay 1 if it meant getting rid of these random errors.