Since commit 23e8b5d8, the code is refactored in a way that supports
domains with multiple graphics elements and commit 37b415200 allows
starting such domains. However none of those commits take migration
into account. Even though qemu doesn't support relocation for
anything else than SPICE and for no more than one graphics, there is no
reason to hardcode one graphics into this part of the code as well.
This flag is meant for errors happening on the source of the migration
and isn't used on the destination. To allow better migration
compatibility, don't propagate it to the destination.
Paolo Bonzini pointed out that it's actually possible to migrate a qemu
instance that was paused due to I/O error and it will be able to work on
the destination if the storage is accessible.
This patch introduces flag VIR_MIGRATE_ABORT_ON_ERROR that cancels the
migration in case an I/O error happens while it's being performed and
allows migration without this flag. This flag can be possibly used for
other error reasons that may be introduced in the future.
Currently, we wait for SPICE to migrate in the very same loop where we
wait for qemu to migrate. This has a disadvantage of slowing seamless
migration down. One one hand, we should not kill the domain until all
SPICE data has been migrated. On the other hand, there is no need to
wait in the very same loop and hence slowing down 'cont' on the
destination. For instance, if users are watching a movie, they can
experience the movie to be stopped for a couple of seconds, as
processors are not running nor on src nor on dst as libvirt waits for
SPICE to migrate. We should move the waiting phase to migration CONFIRM
phase.
During a live migration the guest may receive a disk access I/O error.
In this state the guest is unable to continue running on a remote host
after migration as some state may be present in the kernel and not
migrated.
With this patch, the migration is canceled in such case so it can either
continue on the source if the I/O issues are recovered or has to be
destroyed anyways.
Currently, a listen address for a SPICE server can be specified. Later,
when the domain is migrated, we need to relocate the graphics which
involves telling new destination to the SPICE server. However, we can't
just assume the listen address is the new location, because the listen
address can be ANYCAST (0.0.0.0 for IPv4, :: for IPv6). In which case,
we want to pass the remote hostname. But there are some troubles with
ANYCAST. In both IPv4 and IPv6 it has many ways for specifying such
address. For instance, in IPv4: 0, 0.0, 0.0.0, 0.0.0.0. The number of
variations gets bigger in IPv6 world. Hence, in order to check for
ANYCAST address sanely, we should take the provided listen address,
parse it and format back in it's full form. Which is exactly what this
patch does.
A literal IPv6 must be escaped, otherwise migration fails with:
unable to execute QEMU command 'drive-mirror': address resolution failed
for f0::0d:5901: Servname not supported for ai_socktype
since QEMU treats everything after the first ':' as the port.
Since f03dcc5 we use [::] as the listening address both on qemu
command line in -incoming and in nbd-server-start QMP command.
However the latter requires just :: without the braces.
Commit 7f15ebc7a2 introduced a bug
happening when guests without a <graphics> element are migrated.
The initialization of listenAddress happens unconditionally
from the cookie even if the cookie->graphics pointer was NULL.
Moved the initialization to where it is safe.
Signed-off-by: Viktor Mihajlovski <mihajlov@linux.vnet.ibm.com>
Currently the parsing of XML is pushed down into the various
migration helper APIs. This makes it difficult to insert the
correct access control checks, since one helper API services
many public APIs. Pull the parsing of XML up to the top level
of the QEMU driver APIs
Several APIs allow for custom XML to be passed in. This is
checked for ABI stability, which will ensure the UUID is
not being changed. There isn't validation that the name
did not change though. This could allow renaming of guests
via the backdoor, which in turn could allow for bypassing
access control restrictions based on names.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
Currently the virGetHostname() API has a bogus virConnectPtr
parameter. This is because virtualization drivers directly
reference this API in their virDriverPtr tables, tieing its
API design to the public virConnectGetHostname API design.
This also causes problems for access control checks since
these must only be done for invocations from the public
API, not internal invocation.
Remove the bogus virConnectPtr parameter, and make each
hypervisor driver provide a dedicated function for the
driver API impl. This will allow access control checks
to be easily inserted later.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The QEMU migration code unconditionally sets the 'persistent'
cookie flag on the source host. The dest host, however, only
allows it during parsing if VIR_MIGRATE_PERSIST_DEST was
set. Make the source host only set it if this flag is
present.
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
The source code base needs to be adapted as well. Some files
include virutil.h just for the string related functions (here,
the include is substituted to match the new file), some include
virutil.h without any need (here, the include is removed), and
some require both.
Instead of calling virCgroupForDomain every time we need
the virCgrouPtr instance, just do it once at Vm startup
and cache a reference to the object in qemuDomainObjPrivatePtr
until shutdown of the VM. Removing the virCgroupPtr from
the QEMU driver state also means we don't have stale mount
info, if someone mounts the cgroups filesystem after libvirtd
has been started
Signed-off-by: Daniel P. Berrange <berrange@redhat.com>
When migrating a domain with disk images stored locally (and using
storage migration), we should not complain about unsafe migration no
matter what cache policy is used for that disk.
https://bugzilla.redhat.com/show_bug.cgi?id=920441
Currently, we are discarding listen attribute from qemu cookie even though
we strive to gather it. This result in not so cool bug: if user have
different networks, one for management/migration, and one for VNC/SPICE we
pass incorrect host to the qemu in client_migrate_info. What we actually
pass is remote hostname, while we should be passing remote listen address.
It doesn't matter as long as these two are the same, but they don't need
necessary to be like that.
This patch is the result of running:
for i in $(git ls-files | grep -v html | grep -v \.po$ ); do
sed -i -e "s/virDomainXMLConf/virDomainXMLOption/g" -e "s/xmlconf/xmlopt/g" $i
done
and a few manual tweaks.
Allow migration over IPv6 by listening on [::] instead of 0.0.0.0
when QEMU supports it (QEMU_CAPS_IPV6_MIGRATION) and there is
at least one v6 address configured on the system.
Use virURIParse in qemuMigrationPrepareDirect to allow parsing
IPv6 addresses, which would cause an 'incorrect :port' error
message before.
Move setting of migrateFrom from qemuMigrationPrepare{Direct,Tunnel}
after domain XML parsing, since we need the QEMU binary path from it
to get its capabilities.
Bug: https://bugzilla.redhat.com/show_bug.cgi?id=846013
Since we switched from direct host migration scheme to the one,
where we connect to the destination and then just pass a FD to a
qemu, we have uncovered a qemu bug. Qemu expects migration FD to
block. However, we are passing a nonblocking one which results in
cryptic error messages like:
qemu: warning: error while loading state section id 2
load of migration failed
The bug is already known to Qemu folks, but we should workaround
already released Qemus. Patch has been originally proposed by Stefan
Hajnoczi <stefanha@gmail.com>
The virCaps structure gathered a ton of irrelevant data over time that.
The original reason is that it was propagated to the XML parser
functions.
This patch aims to create a new data structure virDomainXMLConf that
will contain immutable data that are used by the XML parser. This will
allow two things we need:
1) Get rid of the stuff from virCaps
2) Allow us to add callbacks to check and add driver specific stuff
after domain XML is parsed.
This first attempt removes pointers to private data allocation functions
to this new structure and update all callers and function that require
them.
qemuProcessStart expects to be run with a job already set and every
caller except for qemuMigrationPrepareAny use it correctly. This bug can
be observed in libvirtd logs during incoming migration as
warning : qemuDomainObjEnterMonitorInternal:979 : This thread seems
to be the async job owner; entering monitor without asking for a
nested job is dangerous
This function does the source part of NBD magic. It
invokes drive-mirror on each non shared and RW disk with
a source and wait till the mirroring process completes.
When it does we can proceed with migration.
Currently, an active waiting is done: every 500ms libvirt
asks qemu if block-job is finished or not. However, once
the job finishes, qemu doesn't report its progress so we
can only assume if the job finished successfully or not.
The better solution would be to listen to the event which
is sent as soon as the job finishes. The event does
contain the result of job.
We need to start NBD server and feed it with all non-<shared/>,
RW and source-full disks. Moreover, with new virPortAllocator we
must ensure the borrowed port for NBD server will be returned if
either migration completes or qemu process is torn down.
This migration cookie is meant for two purposes. The first is to be sent
in begin phase from source to destination to let it know we support new
implementation of VIR_MIGRATE_NON_SHARED_{DISK,INC} so destination can
start NBD server. Then, the second purpose is, destination can let us
know, on which port the NBD server is running.
As a side effect, this also fixes reporting disk migration process.
It was added to memory migration progress, which was wrong. Disk
progress has dedicated fields in virDomainJobInfo structure.
To avoid having to hold the qemu driver lock while iterating through
close callbacks and calling them. This fixes a real deadlock when a
domain which is being migrated from another host gets autodestoyed as a
result of broken connection to the other host.
Due to "feature"/"features" nasty typo, any features marked as mandatory
by one side of a migration are silently considered optional by the other
side. The following is the code that formats mandatory features in
migration cookie:
for (i = 0 ; i < QEMU_MIGRATION_COOKIE_FLAG_LAST ; i++) {
if (mig->flagsMandatory & (1 << i))
virBufferAsprintf(buf, " <feature name='%s'/>\n",
qemuMigrationCookieFlagTypeToString(i));
}
With the majority of fields in the virQEMUDriverPtr struct
now immutable or self-locking, there is no need for practically
any methods to be using the QEMU driver lock. Only a handful
of helper APIs in qemu_conf.c now need it
