melanie witt db455548a1 Use force=True for os-brick disconnect during delete

The 'force' parameter of os-brick's disconnect_volume() method allows
callers to ignore flushing errors and ensure that devices are being
removed from the host.

We should use force=True when we are going to delete an instance to
avoid leaving leftover devices connected to the compute host which
could then potentially be reused to map to volumes to an instance that
should not have access to those volumes.

We can use force=True even when disconnecting a volume that will not be
deleted on termination because os-brick will always attempt to flush
and disconnect gracefully before forcefully removing devices.

Closes-Bug: #2004555

Change-Id: I3629b84d3255a8fe9d8a7cea8c6131d7c40899e8

2023-05-10 07:09:05 -07:00

13 KiB

Raw Blame History

Live-migrate instances

Live-migrating an instance means moving its virtual machine to a different OpenStack Compute server while the instance continues running. Before starting a live-migration, review the chapter section_configuring-compute-migrations. It covers the configuration settings required to enable live-migration, but also reasons for migrations and non-live-migration options.

The instructions below cover shared-storage and volume-backed migration. To block-migrate instances, add the command-line option -block-migrate to the nova live-migration command, and --block-migration to the openstack server migrate command.

Manual selection of the destination host

Obtain the ID of the instance you want to migrate:

$ openstack server list

+--------------------------------------+------+--------+-----------------+------------+
| ID                                   | Name | Status | Networks        | Image Name |
+--------------------------------------+------+--------+-----------------+------------+
| d1df1b5a-70c4-4fed-98b7-423362f2c47c | vm1  | ACTIVE | private=a.b.c.d | ...        |
| d693db9e-a7cf-45ef-a7c9-b3ecb5f22645 | vm2  | ACTIVE | private=e.f.g.h | ...        |
+--------------------------------------+------+--------+-----------------+------------+

Determine on which host the instance is currently running. In this example, vm1 is running on HostB:

$ openstack server show d1df1b5a-70c4-4fed-98b7-423362f2c47c

+----------------------+--------------------------------------+
| Field                | Value                                |
+----------------------+--------------------------------------+
| ...                  | ...                                  |
| OS-EXT-SRV-ATTR:host | HostB                                |
| ...                  | ...                                  |
| addresses            | a.b.c.d                              |
| flavor               | m1.tiny                              |
| id                   | d1df1b5a-70c4-4fed-98b7-423362f2c47c |
| name                 | vm1                                  |
| status               | ACTIVE                               |
| ...                  | ...                                  |
+----------------------+--------------------------------------+

Select the compute node the instance will be migrated to. In this example, we will migrate the instance to HostC, because nova-compute is running on it:

$ openstack compute service list

+----+------------------+-------+----------+---------+-------+----------------------------+
| ID | Binary           | Host  | Zone     | Status  | State | Updated At                 |
+----+------------------+-------+----------+---------+-------+----------------------------+
|  3 | nova-conductor   | HostA | internal | enabled | up    | 2017-02-18T09:42:29.000000 |
|  4 | nova-scheduler   | HostA | internal | enabled | up    | 2017-02-18T09:42:26.000000 |
|  5 | nova-compute     | HostB | nova     | enabled | up    | 2017-02-18T09:42:29.000000 |
|  6 | nova-compute     | HostC | nova     | enabled | up    | 2017-02-18T09:42:29.000000 |
+----+------------------+-------+----------+---------+-------+----------------------------+

Check that HostC has enough resources for migration:

$ openstack host show HostC

+-------+------------+-----+-----------+---------+
| Host  | Project    | CPU | Memory MB | Disk GB |
+-------+------------+-----+-----------+---------+
| HostC | (total)    |  16 |     32232 |     878 |
| HostC | (used_now) |  22 |     21284 |     422 |
| HostC | (used_max) |  22 |     21284 |     422 |
| HostC | p1         |  22 |     21284 |     422 |
| HostC | p2         |  22 |     21284 |     422 |
+-------+------------+-----+-----------+---------+

cpu: Number of CPUs
memory_mb: Total amount of memory, in MB
disk_gb: Total amount of space for NOVA-INST-DIR/instances, in GB

In this table, the first row shows the total amount of resources available on the physical server. The second line shows the currently used resources. The third line shows the maximum used resources. The fourth line and below shows the resources available for each project.

Migrate the instance:

$ openstack server migrate d1df1b5a-70c4-4fed-98b7-423362f2c47c --live-migration --host HostC

Confirm that the instance has been migrated successfully:

$ openstack server show d1df1b5a-70c4-4fed-98b7-423362f2c47c

+----------------------+--------------------------------------+
| Field                | Value                                |
+----------------------+--------------------------------------+
| ...                  | ...                                  |
| OS-EXT-SRV-ATTR:host | HostC                                |
| ...                  | ...                                  |
+----------------------+--------------------------------------+

If the instance is still running on HostB, the migration failed. The nova-scheduler and nova-conductor log files on the controller and the nova-compute log file on the source compute host can help pin-point the problem.

Automatic selection of the destination host

To leave the selection of the destination host to the Compute service, use the nova command-line client.

Obtain the instance ID as shown in step 1 of the section section-manual-selection-of-dest.
Leave out the host selection steps 2, 3, and 4.

Migrate the instance:

$ nova live-migration d1df1b5a-70c4-4fed-98b7-423362f2c47c

Monitoring the migration

Confirm that the instance is migrating:

$ openstack server show d1df1b5a-70c4-4fed-98b7-423362f2c47c

+----------------------+--------------------------------------+
| Field                | Value                                |
+----------------------+--------------------------------------+
| ...                  | ...                                  |
| status               | MIGRATING                            |
| ...                  | ...                                  |
+----------------------+--------------------------------------+

Check progress

Use the nova command-line client for nova's migration monitoring feature. First, obtain the migration ID:

$ nova server-migration-list d1df1b5a-70c4-4fed-98b7-423362f2c47c
+----+-------------+-----------  (...)
| Id | Source Node | Dest Node | (...)
+----+-------------+-----------+ (...)
| 2  | -           | -         | (...)
+----+-------------+-----------+ (...)

For readability, most output columns were removed. Only the first column, Id, is relevant. In this example, the migration ID is 2. Use this to get the migration status.

$ nova server-migration-show d1df1b5a-70c4-4fed-98b7-423362f2c47c 2
+------------------------+--------------------------------------+
| Property               | Value                                |
+------------------------+--------------------------------------+
| created_at             | 2017-03-08T02:53:06.000000           |
| dest_compute           | controller                           |
| dest_host              | -                                    |
| dest_node              | -                                    |
| disk_processed_bytes   | 0                                    |
| disk_remaining_bytes   | 0                                    |
| disk_total_bytes       | 0                                    |
| id                     | 2                                    |
| memory_processed_bytes | 65502513                             |
| memory_remaining_bytes | 786427904                            |
| memory_total_bytes     | 1091379200                           |
| server_uuid            | d1df1b5a-70c4-4fed-98b7-423362f2c47c |
| source_compute         | compute2                             |
| source_node            | -                                    |
| status                 | running                              |
| updated_at             | 2017-03-08T02:53:47.000000           |
+------------------------+--------------------------------------+

The output shows that the migration is running. Progress is measured by the number of memory bytes that remain to be copied. If this number is not decreasing over time, the migration may be unable to complete, and it may be aborted by the Compute service.

Note

The command reports that no disk bytes are processed, even in the event of block migration.

What to do when the migration times out

During the migration process, the instance may write to a memory page after that page has been copied to the destination. When that happens, the same page has to be copied again. The instance may write to memory pages faster than they can be copied, so that the migration cannot complete. There are two optional actions, controlled by :oslo.configlibvirt.live_migration_timeout_action, which can be taken against a VM after :oslo.configlibvirt.live_migration_completion_timeout is reached:

abort (default): The live migration operation will be cancelled after the completion timeout is reached. This is similar to using API DELETE /servers/{server_id}/migrations/{migration_id}.
force_complete: The compute service will either pause the VM or trigger post-copy depending on if post copy is enabled and available (:oslo.configlibvirt.live_migration_permit_post_copy is set to True). This is similar to using API POST /servers/{server_id}/migrations/{migration_id}/action (force_complete).

You can also read the :oslo.configlibvirt.live_migration_timeout_action configuration option help for more details.

The following remarks assume the KVM/Libvirt hypervisor.

How to know that the migration timed out

To determine that the migration timed out, inspect the nova-compute log file on the source host. The following log entry shows that the migration timed out:

# grep WARNING.*d1df1b5a-70c4-4fed-98b7-423362f2c47c /var/log/nova/nova-compute.log
...
WARNING nova.virt.libvirt.migration [req-...] [instance: ...]
live migration not completed after 1800 sec

Addressing migration timeouts

To stop the migration from putting load on infrastructure resources like network and disks, you may opt to cancel it manually.

$ nova live-migration-abort INSTANCE_ID MIGRATION_ID

To make live-migration succeed, you have several options:

Manually force-complete the migration
```
$ nova live-migration-force-complete INSTANCE_ID MIGRATION_ID
```
The instance is paused until memory copy completes.

Caution

Since the pause impacts time keeping on the instance and not all applications tolerate incorrect time settings, use this approach with caution.
Enable auto-convergence

Auto-convergence is a Libvirt feature. Libvirt detects that the migration is unlikely to complete and slows down its CPU until the memory copy process is faster than the instance's memory writes.

To enable auto-convergence, set live_migration_permit_auto_converge=true in nova.conf and restart nova-compute. Do this on all compute hosts.

Caution

One possible downside of auto-convergence is the slowing down of the instance.
Enable post-copy

This is a Libvirt feature. Libvirt detects that the migration does not progress and responds by activating the virtual machine on the destination host before all its memory has been copied. Access to missing memory pages result in page faults that are satisfied from the source host.

To enable post-copy, set live_migration_permit_post_copy=true in nova.conf and restart nova-compute. Do this on all compute hosts.

When post-copy is enabled, manual force-completion does not pause the instance but switches to the post-copy process.
Caution

Possible downsides:
- When the network connection between source and destination is interrupted, page faults cannot be resolved anymore, and the virtual machine is rebooted.
- Post-copy may lead to an increased page fault rate during migration, which can slow the instance down.

If live migrations routinely timeout or fail during cleanup operations due to the user token timing out, consider configuring nova to use service user tokens <service_user_token>.

13 KiB Raw Blame History