Linux Network: NIC team problem in CentOS7

I have a computer interface with a single physical NIC em1 attached. Then I try to add another NIC em2 to form a team with the following command:

Teamdctl Team0 port add em2

After adding, the second interface of the NIC device starts to work incorrectly: there is a 50% packet loss and there is no network activity in the newly added em2 Interface. I have verified that the link is up. em2 Y
I send Teamdctl Team0 port add em2 shows that everything is fine

                prepare:
Runner: roundrobin
kernel equipment mode: roundrobin
D-BUS enabled: yes
ZeroMQ enabled: no
Debug level: 0
demonized: no
PID: 29838
PID file: /var/run/teamd/team0.pid
ports:
em1
ifindex: 2
addr: 78: 2b: cb: 20: faith: 66
Ethtool link: 1000mbit / fullduplex / up
Link watches:
link summary: top
example[link_watch_0]:
name: ethtool
link up
countdown: 0
Link delay: 0
link down delay: 0
em2
ifindex: 3
addr: 78: 2b: cb: 20: faith: 66
Ethtool link: 1000mbit / fullduplex / up
Link watches:
link summary: top
example[link_watch_0]:
name: ethtool
link up
countdown: 0
Link delay: 0
link down delay: 0

It works correctly only after restarting the network with systemctl restart the network I send.
But it is essential to add a second NIC to the computer without restarting the network. Is it possible in CentOS 7 with teamd 1.27 installed?

It can be important:
The interface of team0 is linked to the interface of bridge0 as follows:

em1 | -------- |
|
| ---------- | team0 | -------- | bridge0 (IPv4 addr)
|
em2 | -------- |

Is it possible that the bridge is the cause of the problem?