sls_updater.py
Technical DetailsNo action needed. Informational purposes only.
This migration is performed offline for data security. The running SLS file is first dumped, then the migration script is run and a new, migrated output file is created.
leaf
to leaf-bmc
, and agg
to leaf
.api-gateway
entries from HMLB subnets for CSM 1.2 security.kubeapi-vip
reservations for all networks except NMN.uai_macvlan
in NMN DHCP ranges and uai_macvlan
VLAN.--retain-unused-user-network
to keep).Switch names change in CSM 1.2 and must be applied in the following order:
leaf
switches become leaf-bmc
switches.agg
switches become leaf
switches.This needs to be done in the order listed above.
api-gateway
/ istio-ingress-gateway
reservations from HMNLB subnetsFor CSM 1.2, the API gateway no longer listens on the HMNLB MetalLB address pool. These aliases provided DNS records and have been removed in CSM 1.2.
New for CSM 1.2: The BICAN network ExtraProperties
value of SystemDefaultRoute
is used
to point to the CAN, CHN, or CMN, and is used by utilities to systematically toggle routes.
Using the existing CAN as a template, create the CMN. The same IP addresses will be preserved for
NCNs (bootstrap_dhcp
). A new network_hardware
subnet will be created where the end of the previous
bootstrap_dhcp
subnet existed to contain switching hardware. MetalLB pools in the bootstrap_dhcp
subnet will be shifted around to remain at the end of the new bootstrap subnet.
With the original CAN as a template, the new CHN will be created. IP addresses will come from the
--customer-highspeed-network <vlan> <ipaddress>
(or its defaults). This is be created by default, but
can be removed (if not needed or desired) by using the --retain-unused-user-network
flag.
Since the original/existing CAN has been converted to the new CMN, the CAN must have new IP addresses.
These are provided using the --customer-access-network <vlan> <ipaddress>
(or its defaults). This CAN
conversion will happen by default, but the new CAN may be removed (if not needed or desired) by using the
--retain-unused-user-network
flag.
MetalLB and switches now obtain BGP peers using SLS data.
--bgp-asn INTEGER RANGE The autonomous system number for BGP router
[default: 65533;64512<=x<=65534]
--bgp-cmn-asn INTEGER RANGE The autonomous system number for CMN BGP
clients [default: 65534;64512<=x<=65534]
--bgp-nmn-asn INTEGER RANGE The autonomous system number for NMN BGP
clients [default: 65533;64512<=x<=65534]
In CMN and NMN:
"Type": "ethernet",
"ExtraProperties": {
"CIDR": "10.102.3.0/25",
"MTU": 9000,
"MyASN": 65536,
"PeerASN": 65533,
"Subnets":
kubeapi-vip
reservations for all networks except NMNSelf explanatory. This endpoint now exists only on the NMN.
uai_macvlan
in NMN ranges and uai_macvlan
VLANSelf explanatory. Ranges are used for the addresses of UAIs.
By default, the CAN will be removed if --bican-user-network-name CHN
is specified, or the CHN will be removed if
--bican-user-network-name CAN
is specified. In order to keep a network from being removed, use the --retain-unused-user-network
flag.
Retention of the unused network is not normal behavior.
Go back to SLS Upgrade.