diff --git a/docs/book/known_issues.md b/docs/book/known_issues.md index c6a462a6b1..f5e122d87c 100644 --- a/docs/book/known_issues.md +++ b/docs/book/known_issues.md @@ -56,3 +56,37 @@ Note: This Issue is present in all Kubernetes Releases - Impact: If a user tries to expand a PVC to a size which may not be supported by the underlying storage system, volume expansion will keep failing and there is no way to recover. - Issue: https://github.com/kubernetes/enhancements/issues/1790 - Workaround is being tracked at https://github.com/kubernetes/enhancements/blob/master/keps/sig-storage/1790-recover-resize-failure/README.md + +## Multipath and VMware Virtual Disks + +When following generic multipath setup guides (e.g. [NetApp Trident worker node prep](https://docs.netapp.com/us-en/trident/trident-use/worker-node-prep.html)), multipath may be enabled for all disks, including VMware virtual disks. This can cause issues with the vsphere-csi-driver: + +- VMware virtual disks are presented as device-mapper devices (e.g. `/dev/dm-*`). +- The vsphere-csi-driver expects a physical device, not a virtual device. +- This results in errors like: `/sys/devices/virtual/block/dm-/device: no such file or directory` + +**Root cause:** Multipath is configured for VMware virtual disks, but multipathing should only be used for iSCSI/NFS (e.g. with Trident). For VMware, multipathing is handled at the ESXi layer. + +**Solution:** +Blacklist VMware virtual disks in your `multipath.conf` to prevent multipath from claiming them: + +```conf +defaults { ... } +blacklist { + device { + vendor "VMware" + product "Virtual disk" + } +} +``` + +You can verify the vendor and product with: + +```shell +dmsetup ls --tree +# mpathz (250:99) +# └─ (34:128) + +multipath -l mpathz +# mpathz (56100d29f0cf860b171204fb181814bbd) dm-99 VMware,Virtual disk +```