Guidelines and Limitations for Downgrading
See the following guidelines before downgrading:
-
There is no official Zero Downtime Downgrade support for clustering—However, in some cases, Zero Downtime Downgrading will work. See the following known issues for downgrading; note that there may be other issues that require you to reload your cluster units, thus causing downtime.
-
Downgrade to a pre-9.9(1) release with clustering—9.9(1) and later includes an improvement in the backup distribution. If you have 3 or more units in the cluster, you must perform the following steps:
-
Remove all secondary units from the cluster (so the cluster consists only of the primary unit).
-
Downgrade 1 secondary unit, and rejoin it to the cluster.
-
Disable clustering on the primary unit; downgrade it, and rejoin the cluster.
-
Downgrade the remaining secondary units, and join them back to the cluster, one at a time.
-
-
Downgrade to a pre-9.9(1) release when you enable cluster site redundancy—You should disable site redundancy if you want to downgrade (or if you want to add a pre-9.9(1) unit to a cluster). Otherwise, you will see side effects, for example, dummy forwarding flows on the unit running the old version.
-
Downgrade from 9.8(1) with clustering and crypto-map—There is no Zero Downtime Downgrade support when downgrading from 9.8(1) when you have a crypto-map configured. You should clear the crypto-map configuration before downgrading, and then re-apply the configuration after the downgrade.
-
Downgrade from 9.8(1) with clustering unit health check set to .3 to .7 seconds—If you downgrade your ASA software after setting the hold time to .3 - .7 (health-check holdtime ), this setting will revert to the default of 3 seconds because the new setting is unsupported.
-
Downgrade from 9.5(2) or later to 9.5(1) or earlier with clustering (CSCuv82933)—There is no Zero Downtime Downgrade support when downgrading from 9.5(2). You must reload all units at roughly the same time so that a new cluster is formed when the units come back online. If you wait to reload the units sequentially, then they will be unable to form a cluster.
-
Downgrade from 9.2(1) or later to 9.1 or earlier with clustering—Zero Downtime Downgrade is not supported.
-
-
Downgrade issue from 9.22 or later—If you disable the USB port using the usb-port disable command, but then downgrade to an earlier release, the port will remain disabled, and you cannot re-enable it without erasing the NVRAM (the FXOS local-mgmt erase secure all command).
-
Downgrade issue from 9.18 or later—There is a behavior change in 9.18 where the access-group command will be listed before its access-list commands. If you downgrade, the access-group command will be rejected because it has not yet loaded the access-list commands. This outcome occurs even if you had previously enabled the forward-reference enable command, because that command is now removed. Before you downgrade, be sure to copy all access-group commands manually, and then after downgrading, re-enter them.
-
Downgrade issue for the Firepower 2100 in Platform mode from 9.13/9.14 to 9.12 or earlier—For a Firepower 2100 with a fresh installation of 9.13 or 9.14 that you converted to Platform mode: If you downgrade to 9.12 or earlier, you will not be able to configure new interfaces or edit existing interfaces in FXOS (note that 9.12 and earlier only supports Platform mode). You either need to restore your version to 9.13 or later, or you need to clear your configuration using the FXOS erase configuration command. This problem does not occur if you originally upgraded to 9.13 or 9.14 from an earlier release; only fresh installations are affected, such as a new device or a re-imaged device. (CSCvr19755)
-
Downgrade from 9.10(1) for smart licensing—Due to changes in the smart agent, if you downgrade, you must re-register your device to the Cisco Smart Software Manager. The new smart agent uses an encrypted file, so you need to re-register to use an unencrypted file required by the old smart agent.
-
Downgrade to 9.5 and earlier with passwords using PBKDF2 (Password-Based Key Derivation Function 2) hash—Versions before 9.6 do not support PBKDF2 hashing. In 9.6(1), enable and username passwords longer than 32 characters use PBKDF2 hashing. In 9.7(1), new passwords of all lengths use PBKDF2 hashing (existing passwords continue to use MD5 hashing). If you downgrade, the enable password reverts to the default (which is blank). Usernames will not parse correctly, and the username commands will be removed. You must re-create your local users.
-
Downgrade from Version 9.5(2.200) for the ASA Virtual—The ASA virtual does not retain the licensing registration state. You need to re-register with the license smart register idtoken id_token force command (for ASDM: see the Configuration > Device Management > Licensing > Smart Licensing page, and use the Force registration option); obtain the ID token from the Smart Software Manager.
-
VPN tunnels are replicated to the standby unit even if the standby unit is running a version of software that does not support the Ciphersuite that the original tunnel negotiated—This scenario occurs when downgrading. In this case, disconnect your VPN connection and reconnect.