2015-05-21 20:17:54 |
Vladimir Kozhukalov |
bug |
|
|
added bug |
2015-05-21 20:19:02 |
Mike Scherbakov |
tags |
fuel-docs |
docs |
|
2015-05-21 20:20:33 |
Vladimir Kozhukalov |
description |
We need to mention this in our documentation.
As far as we are building target OS images for Ubuntu on the master node, and it requires quite high-performance disk subsystem, it is strictly recommended for KVM users to use cache='unsafe' attribute in their libvirt disks definitions when they use KVM/QEMU VM for their master node.
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2' cache='unsafe' io='native'/>
<source file='/var/lib/libvirt/images/fuel-master.qcow2'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/>
</disk> |
We need to mention this in our documentation, otherwise building image could take more than an hour and so deployment could fail due to provision timeout.
As far as we are building target OS images for Ubuntu on the master node, and it requires quite high-performance disk subsystem, it is strictly recommended for KVM users to use cache='unsafe' attribute in their libvirt disks definitions when they use KVM/QEMU VM for their master node.
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2' cache='unsafe' io='native'/>
<source file='/var/lib/libvirt/images/fuel-master.qcow2'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/>
</disk> |
|
2015-05-21 20:21:46 |
Vladimir Kozhukalov |
description |
We need to mention this in our documentation, otherwise building image could take more than an hour and so deployment could fail due to provision timeout.
As far as we are building target OS images for Ubuntu on the master node, and it requires quite high-performance disk subsystem, it is strictly recommended for KVM users to use cache='unsafe' attribute in their libvirt disks definitions when they use KVM/QEMU VM for their master node.
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2' cache='unsafe' io='native'/>
<source file='/var/lib/libvirt/images/fuel-master.qcow2'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/>
</disk> |
We need to mention this in our documentation.
As far as we are building target OS images for Ubuntu on the master node, and it requires quite high-performance disk subsystem, it is strictly recommended for KVM users to use cache='unsafe' attribute in their libvirt disks definitions when they use KVM/QEMU VM for their master node. Otherwise, building target OS image could take more than an hour and so the deployment could fail due to provision timeout.
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2' cache='unsafe' io='native'/>
<source file='/var/lib/libvirt/images/fuel-master.qcow2'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/>
</disk> |
|
2015-07-01 11:50:46 |
Evgeny Konstantinov |
fuel: assignee |
Fuel Documentation Team (fuel-docs) |
Evgeny Konstantinov (evkonstantinov) |
|
2015-07-01 12:27:40 |
OpenStack Infra |
fuel: status |
Triaged |
In Progress |
|
2015-07-03 07:05:09 |
Alexei Sheplyakov |
summary |
Use cache='unsafe' for KVM based master node |
IBP images build is way too slow |
|
2015-07-03 07:15:35 |
Alexei Sheplyakov |
tags |
docs |
|
|
2015-07-03 07:15:57 |
Alexei Sheplyakov |
fuel: assignee |
Evgeny Konstantinov (evkonstantinov) |
Alexei Sheplyakov (asheplyakov) |
|
2015-07-03 07:16:03 |
Alexei Sheplyakov |
fuel: status |
In Progress |
Confirmed |
|
2015-07-11 17:43:42 |
Mike Scherbakov |
nominated for series |
|
fuel/7.0.x |
|
2015-07-11 17:43:42 |
Mike Scherbakov |
bug task added |
|
fuel/7.0.x |
|
2015-07-11 17:43:42 |
Mike Scherbakov |
nominated for series |
|
fuel/6.1.x |
|
2015-07-11 17:43:42 |
Mike Scherbakov |
bug task added |
|
fuel/6.1.x |
|
2015-07-11 17:44:15 |
Mike Scherbakov |
fuel/6.1.x: milestone |
|
6.1-updates |
|
2015-07-11 17:44:22 |
Mike Scherbakov |
fuel/7.0.x: milestone |
6.1 |
7.0 |
|
2015-07-13 07:55:35 |
Alexei Sheplyakov |
description |
We need to mention this in our documentation.
As far as we are building target OS images for Ubuntu on the master node, and it requires quite high-performance disk subsystem, it is strictly recommended for KVM users to use cache='unsafe' attribute in their libvirt disks definitions when they use KVM/QEMU VM for their master node. Otherwise, building target OS image could take more than an hour and so the deployment could fail due to provision timeout.
<disk type='file' device='disk'>
<driver name='qemu' type='qcow2' cache='unsafe' io='native'/>
<source file='/var/lib/libvirt/images/fuel-master.qcow2'/>
<target dev='vda' bus='virtio'/>
<address type='pci' domain='0x0000' bus='0x00' slot='0x09' function='0x0'/>
</disk> |
Due to a very suboptimal IO layering building the target OS (Ubuntu)
IBP images takes an unreasonably long time (more than an hour!) if
the master node runs in the KVM VM.
fuel_agent (which builds Ubuntu images) runs in the mcollective
container. A container filesystem is backed by dm-snapshot which
is known to have a poor write performance, especially for large
files (like OS images). To add an insult to injury dm-snapshot
storage is a loopback file residing in /var/lib/docker.
Thus the data written to the OS image being built goes through
a number of layers:
ext4 -> loop -> dm-snapshot (/var/lib/docker) -> loop ->
-> ext4 (/var/lib/docker) -> dm-linear (os-varlibdocker) ->
-> hard drive
If the master node runs on a VM rather than a physcial hardware
the above chain gets even more complicated (and write performance
gets really awful)
ext4 -> loop -> dm-snapshot (/var/lib/docker) -> loop ->
-> ext4 (/var/lib/docker) -> dm-linear (os-varlibdocker) ->
-> virtual hard drive (qemu) -> qcow2 -> ext4 (host)
We need to remove the extra layers to make IO performance reasonable. |
|
2015-07-13 09:57:07 |
Aleksandr Didenko |
fuel/7.0.x: status |
Confirmed |
Fix Committed |
|
2015-07-13 09:57:42 |
Aleksandr Didenko |
fuel/6.1.x: assignee |
|
MOS Sustaining (mos-sustaining) |
|
2015-07-13 09:57:47 |
Aleksandr Didenko |
fuel/6.1.x: importance |
Undecided |
High |
|
2015-07-13 09:57:51 |
Aleksandr Didenko |
fuel/6.1.x: status |
New |
Confirmed |
|
2015-07-20 21:03:55 |
Vitaly Sedelnik |
fuel/6.1.x: status |
Confirmed |
Won't Fix |
|
2015-09-28 10:26:40 |
Maksym Strukov |
fuel/7.0.x: status |
Fix Committed |
Fix Released |
|
2015-09-28 10:26:46 |
Maksym Strukov |
fuel: status |
Fix Committed |
Fix Released |
|