yahoo-eng-team team mailing list archive
-
yahoo-eng-team team
-
Mailing list archive
-
Message #95110
[Bug 2092529] [NEW] Volume Size Mismatch in PowerFlex (ScaleIO) When Extending Multiple Volumes
Public bug reported:
Description
===========
When simultaneously extending additional volumes attached to separate running VMs in a PowerFlex environment, a mismatch occurs between the physical size and capacity size of the extended volumes. This issue resolves only after rebooting the VM.
Sequential extensions with a 5-second interval avoid this problem, but
rebooting is not an acceptable workaround for HA-configured VMs.
Steps to reproduce
==================
1. Create and attach additional volumes(powerflex) to multiple running VMs.
2. Simultaneously extend these volumes.
3. Verify volume sizes on the compute host using:
$ virsh domblkinfo <instance_id> <device_path>
Expected Behavior
=================
The capacity and physical sizes should match immediately after the extension without requiring a VM reboot.
Actual result
===============
The physical size differs from the capacity size. Example output:
$ virsh domblkinfo 853 vdb --human
Capacity: 16.000 GiB
Allocation: 0.000 B
Physical: 32.000 GiB
Workaround
===============
- Perform volume extensions sequentially with a 5-second interval.
- Reboot the VM to resolve the mismatch.
Environment
===========
- OpenStack Version: Caracal 24.01
- Hypervisor: Libvirt 10.7.0 + QEMU 9.1.0
Logs
=====
Logs from nova during the volume extension operation:
2024-12-26 06:09:11.897 Extend ScaleIO Volume /dev/disk/by-id/emc-vol-xxx-bff4d4b0000005a; new_size=32G
2024-12-26 06:09:12.072 Extend ScaleIO Volume /dev/disk/by-id/emc-vol-xxx-bff4d4c0000005e; new_size=16G
2024-12-26 06:09:12.386 Extend ScaleIO Volume /dev/disk/by-id/emc-vol-xxx-bff4d4d00000060; new_size=32G
** Affects: nova
Importance: Undecided
Status: New
** Summary changed:
- Volume Extension Size Mismatch in PowerFlex
+ Volume Size Mismatch in PowerFlex (ScaleIO) When Extending Multiple Volumes
--
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/2092529
Title:
Volume Size Mismatch in PowerFlex (ScaleIO) When Extending Multiple
Volumes
Status in OpenStack Compute (nova):
New
Bug description:
Description
===========
When simultaneously extending additional volumes attached to separate running VMs in a PowerFlex environment, a mismatch occurs between the physical size and capacity size of the extended volumes. This issue resolves only after rebooting the VM.
Sequential extensions with a 5-second interval avoid this problem, but
rebooting is not an acceptable workaround for HA-configured VMs.
Steps to reproduce
==================
1. Create and attach additional volumes(powerflex) to multiple running VMs.
2. Simultaneously extend these volumes.
3. Verify volume sizes on the compute host using:
$ virsh domblkinfo <instance_id> <device_path>
Expected Behavior
=================
The capacity and physical sizes should match immediately after the extension without requiring a VM reboot.
Actual result
===============
The physical size differs from the capacity size. Example output:
$ virsh domblkinfo 853 vdb --human
Capacity: 16.000 GiB
Allocation: 0.000 B
Physical: 32.000 GiB
Workaround
===============
- Perform volume extensions sequentially with a 5-second interval.
- Reboot the VM to resolve the mismatch.
Environment
===========
- OpenStack Version: Caracal 24.01
- Hypervisor: Libvirt 10.7.0 + QEMU 9.1.0
Logs
=====
Logs from nova during the volume extension operation:
2024-12-26 06:09:11.897 Extend ScaleIO Volume /dev/disk/by-id/emc-vol-xxx-bff4d4b0000005a; new_size=32G
2024-12-26 06:09:12.072 Extend ScaleIO Volume /dev/disk/by-id/emc-vol-xxx-bff4d4c0000005e; new_size=16G
2024-12-26 06:09:12.386 Extend ScaleIO Volume /dev/disk/by-id/emc-vol-xxx-bff4d4d00000060; new_size=32G
To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/2092529/+subscriptions