Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Stage2 #68

Merged
merged 11 commits into from
Apr 18, 2024
Merged

Stage2 #68

merged 11 commits into from
Apr 18, 2024

Conversation

kailiangz1
Copy link
Collaborator

No description provided.

When probe, we can skip suspend when HA, becasue HA
is holding device fd, controller is running and we can
proceed.

Signed-off-by: Kailiang <kailiangz@nvidia.com>
When lm, we can compare the state of vhost which is get
from qemu and the state of controller, if there is no difference,
we can go on without any interrupt.

Signed-off-by: Kailiang <kailiangz@nvidia.com>
Call virtio compare when device config, this way
can cut the downtime of traffic

Signed-off-by: Kailiang <kailiangz@nvidia.com>
Move close vhost after probe can save the time cost of probe
and let the whole downtime low.

Signed-off-by: Kailiang <kailiangz@nvidia.com>
For virtio-blk, if controller is freezed, the pending byte is 0,
in order to fix this, always use 8k buffer which is enough for
current state size

Signed-off-by: Kailiang <kailiangz@nvidia.com>
When probe, there is possiblity backend controller's state
is error, we should reset no matter is HA or not.

Signed-off-by: Kailiang <kailiangz@nvidia.com>
If run into presetup, this means live migration happened after
VFE restart, we can fall back to stage 1 and let the compare
work done by controller side

Signed-off-by: Kailiang <kailiangz@nvidia.com>
After interrupts alloc, traffic will stopped, so, move to
bottome as much as possible

Signed-off-by: Kailiang <kailiangz@nvidia.com>
When live migration, it is useful to log the time to debug
latency issues.

Signed-off-by: Kailiang <kailiangz@nvidia.com>
Stage2 is enabled by default, we can use "--stage1" to
fall back to stage1, debug purpose only.

Signed-off-by: Kailiang <kailiangz@nvidia.com>
Sleep can increase whole downtime, remove to shorten the
whole downtime

Signed-off-by: Kailiang <kailiangz@nvidia.com>
@kailiangz1 kailiangz1 merged commit 4111527 into Mellanox:main Apr 18, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
1 participant