I'm trying to install Ubuntu 26.04.1 LTS on an older HP ProLiant ML310e Gen8 V2, but the installer fails partway through with a warning that the filesystem has been remounted read-only. After that, commands are discarded. I disabled RAID, enabled AHCI, and set USB boot as the first boot option. I also removed the RAID backplane and tried another SATA port, replaced the SATA cables and RAM, and tested with an SSD, but the behavior did not change. Linux Mint XFCE installs successfully, yet the machine reports a critical filesystem error and becomes read-only after booting, usually within a few restarts. The server previously ran Windows Server without any obvious issues. Are there BIOS, storage-controller, or boot settings specific to this ProLiant that need to be changed, or is there another likely cause?
2 Answers
A Linux filesystem normally becomes read-only because the kernel detected storage I/O errors, not because Linux simply dislikes the server. Check the kernel log with `dmesg` or `journalctl -k` and look for ATA errors, failed reads, UNC or medium errors, link resets, or controller timeouts. Those messages will help identify whether the problem is the disk, SATA path, controller, or backplane. Also check the drive’s SMART data and run a full read or extended self-test. Windows working does not completely rule out hardware trouble, since different drivers and workloads can expose different failures.
This model can be temperamental about how its storage bays and controller are configured. One workaround reported for similar systems was installing Linux with the drives in AHCI mode and the front bays disconnected, then reconnecting the drives after the system had booted. Another possibility is using an internal SATA-connected SSD rather than the front-bay controller, although that still needs to be tested for I/O errors. Keep RAID disabled if you do not need it, and verify that the installer is booting in the same UEFI or legacy mode you intend to use for the installed system. If Mint also becomes read-only after a few boots, collect the kernel log before rebooting; repeated corruption points more strongly to a storage-path or controller issue than to the desktop environment.
I’m not using RAID and plan to leave the controller in AHCI mode. Mint can install, but the filesystem becomes corrupted after the second or third boot, so I’ll test the internal SATA connection and inspect the logs before trying again.
I’ll check the kernel logs while trying another distribution. Windows 10 installs and runs normally, so I was starting to wonder whether the machine had some Linux-specific compatibility problem.