| Message ID | 20211126153154.25424-3-lizhang@suse.de (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | migration: multifd live migration improvement | expand |
Li Zhang <lizhang@suse.de> wrote: > When creating the guest on destination with -incoming ip:port in QEMU command line, > the source code that sets the backlog number as the same as multifd channels doesn't > get called. So the number of backlog is always 1. It's very easy to cause live migration > failure, so a bigger number is preferred to reduce the chance of the failure. > > Signed-off-by: Li Zhang <lizhang@suse.de> > --- > migration/socket.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/migration/socket.c b/migration/socket.c > index 05705a32d8..398d4c10fa 100644 > --- a/migration/socket.c > +++ b/migration/socket.c > @@ -152,7 +152,7 @@ socket_start_incoming_migration_internal(SocketAddress *saddr, > QIONetListener *listener = qio_net_listener_new(); > MigrationIncomingState *mis = migration_incoming_get_current(); > size_t i; > - int num = 1; > + int num = 16; > > qio_net_listener_set_name(listener, "migration-socket-listener"); Here, the right answer is to use -incoming defer. Later, Juan.
On 11/26/21 5:32 PM, Juan Quintela wrote: > Li Zhang <lizhang@suse.de> wrote: >> When creating the guest on destination with -incoming ip:port in QEMU command line, >> the source code that sets the backlog number as the same as multifd channels doesn't >> get called. So the number of backlog is always 1. It's very easy to cause live migration >> failure, so a bigger number is preferred to reduce the chance of the failure. >> >> Signed-off-by: Li Zhang <lizhang@suse.de> >> --- >> migration/socket.c | 2 +- >> 1 file changed, 1 insertion(+), 1 deletion(-) >> >> diff --git a/migration/socket.c b/migration/socket.c >> index 05705a32d8..398d4c10fa 100644 >> --- a/migration/socket.c >> +++ b/migration/socket.c >> @@ -152,7 +152,7 @@ socket_start_incoming_migration_internal(SocketAddress *saddr, >> QIONetListener *listener = qio_net_listener_new(); >> MigrationIncomingState *mis = migration_incoming_get_current(); >> size_t i; >> - int num = 1; >> + int num = 16; >> >> qio_net_listener_set_name(listener, "migration-socket-listener"); > Here, the right answer is to use -incoming defer. Ok, thanks a lot. > > Later, Juan. > >
diff --git a/migration/socket.c b/migration/socket.c index 05705a32d8..398d4c10fa 100644 --- a/migration/socket.c +++ b/migration/socket.c @@ -152,7 +152,7 @@ socket_start_incoming_migration_internal(SocketAddress *saddr, QIONetListener *listener = qio_net_listener_new(); MigrationIncomingState *mis = migration_incoming_get_current(); size_t i; - int num = 1; + int num = 16; qio_net_listener_set_name(listener, "migration-socket-listener");
When creating the guest on destination with -incoming ip:port in QEMU command line, the source code that sets the backlog number as the same as multifd channels doesn't get called. So the number of backlog is always 1. It's very easy to cause live migration failure, so a bigger number is preferred to reduce the chance of the failure. Signed-off-by: Li Zhang <lizhang@suse.de> --- migration/socket.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)