CentOS服务器挂载/sysroot失败（initramfs损坏）及KVM VPS崩溃求助

Hey there, let's tackle these two CentOS/KVM issues head-on—here's what you can do to resolve them:

That "bad initramfs" error almost always means your initial RAM filesystem image is corrupted or mismatched with your kernel, which the system needs to boot and mount critical filesystems like /sysroot. Here's how to fix it:

• Boot into Rescue Mode
  Grab a CentOS installation ISO that matches your server's major version, boot from it, and select "Rescue a CentOS system" from the boot menu. Follow the prompts to mount your existing root partition (it'll usually be mounted to /mnt/sysimage).

• Chroot into your system
  Once you're in rescue mode, run this command to access your server's actual operating system environment:

  chroot /mnt/sysimage
  

• Re-generate the initramfs image
  First, list all installed kernel versions to pick the right one:

  rpm -qa | grep kernel
  

  Then re-build the initramfs for your active kernel (replace [kernel-version] with the version string from the output above, e.g., 3.10.0-1160.el7.x86_64):

  dracut -f /boot/initramfs-[kernel-version].img [kernel-version]
  

  If you're not sure which kernel is active, just use the latest one listed—it's the most likely candidate.

• Verify and reboot
  Exit the chroot with exit, then reboot the server. If the corrupted initramfs was the problem, your system should boot normally and mount /sysroot without errors.

• Bonus: Check for filesystem corruption
  If re-generating initramfs doesn't work, your root filesystem might be damaged. Before chrooting in rescue mode, run a filesystem check (make sure the partition is unmounted first!):

  fsck /dev/[your-root-partition]
  

  This can fix underlying filesystem issues that might have corrupted the initramfs in the first place.

Since your VMware instances are rock-solid but the KVM ones crash up to 5 times a month, the issue is almost certainly tied to KVM-specific configurations, storage stack quirks, or system compatibility. Let's break down the troubleshooting steps:

• Check the KVM hypervisor logs first
  Even if you don't have journalctl logs on the VMs themselves, the KVM host's logs will have critical details. Look for entries related to the crashing VMs in:

  • /var/log/libvirt/qemu/[vm-name].log (per-VM QEMU runtime logs)
  • /var/log/messages or /var/log/syslog (host system logs)
    Keep an eye out for errors like IO timeouts, memory allocation failures, or virtio driver crashes—these are common culprits.

• Compare KVM vs VMware VM configurations
  List out the key differences between your KVM and VMware setups:

  • CPU/memory settings (e.g., CPU pinning, memory ballooning, hugepage usage)
  • Disk setup (virtio vs IDE/SCSI drivers, disk cache mode, SSD passthrough vs virtual disks)
  • Network driver (virtio-net vs e1000)
    Misconfigured virtio drivers or disk cache modes (like writethrough which can throttle SSD performance) often cause instability on KVM. Try switching the disk cache mode to writeback or none, and ensure the latest virtio drivers are installed on your CentOS VMs.

• Test SSD performance and health
  Since these are SSD hosts, storage issues could be the root cause. When a KVM VM is up and running, run this fio command to test random write performance (a common pain point for SSDs):

  fio --name=randwrite --ioengine=libaio --rw=randwrite --bs=4k --numjobs=8 --size=1G --iodepth=64 --runtime=60 --time_based
  

  If you see abnormally high latency or IO errors, it might point to hardware degradation or storage contention on the KVM host. Also, check SSD health with smartctl:

  smartctl -a /dev/[your-ssd-device]
  

  Look for attributes like Reallocated_Sector_Ct or Media_Wearout_Indicator—spikes here mean your SSD might be failing.

• Rule out snapshot-related issues
  You're using snapshots of other CentOS machines to fix crashes, but it's possible those snapshots have configurations incompatible with your KVM environment, or repeated restores are corrupting filesystem states. Try deploying a fresh, clean CentOS installation on one of the KVM hosts (not from a snapshot) and monitor its stability. If it doesn't crash, the issue is with your snapshot image, not the KVM infrastructure.

• Update kernels and KVM-related packages
  Outdated kernels often have known bugs with KVM's virtio drivers. On your CentOS VMs, run these commands to update the kernel and related packages:

  yum update kernel kernel-devel virtio*
  

  Reboot the VM after updating, then keep an eye on its stability—this fixes many compatibility issues.

内容的提问来源于stack exchange，提问作者elmazzun