Age | Commit message (Collapse) | Author |
|
Added health-check to sync-group in CLI
Don't use instance health-check when instance in sync group member
Disallow wrong healtch-check configurations
New smoke test
|
|
(cherry picked from commit 3480d92a8c4d84e8c1f94a9362bac2be0cc77921)
|
|
AgentX does not work stable. From time to time we see the system
service crashing/degrading if something is wrong with SNMP from
util net-snmp.
We should disable it by default and enable it only if configured.
set high-availability vrrp snmp
(cherry picked from commit 47875457cd8b176f7f23a3141175d745aeb14d8a)
|
|
By default we enable `--snmp` for keepalived unit service
Add ability to disable it
set high-availability vrrp disable-snmp
(cherry picked from commit 5ae730a52de2f284e45cd433bb0cf66c8508f2f7)
|
|
Fixed setting vrrp version
|
|
Migrate:
high-availability virtual-server 203.0.113.1
to: high-availability virtual-server <name> address 203.0.113.1
|
|
set high-availability vrrp group <name> version 2|3
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Allow multiple ports for high-availability virtual-server
The current implementation allows balance only one "virtual" address
and port between between several "real servers"
Allow matching "fwmark" to set traffic which should be balanced
Allow to set port 0 (all traffic) if we use "fwmark"
Add health-check script
set high-availability virtual-server 203.0.113.1 fwmark '111'
set high-availability virtual-server 203.0.113.1 real-server 192.0.2.11 health-check script '/bin/true'
set high-availability virtual-server 203.0.113.1 real-server 192.0.2.11 port '0'
|
|
When applying the same VRID for IPv4 and IPv6 with RFC3768
compatibility enabled, the IPv6 interfaces came back with the
wrong name. For example:
Name Interface VRID State Priority Last Transition
------ ----------- ------ ------- ---------- -----------------
v4-10 eth1v10 10 MASTER 100 21s
v6-10 vrrpv10 10 MASTER 100 21s
Because of this, the IPv6 interface didn't show up in `show int`.
This change suffixes the interface with the IP version so
`show int` works again.
Name Interface VRID State Priority Last Transition
------ ----------- ------ ------- ---------- -----------------
v4-10 eth1v10v4 10 MASTER 100 21s
v6-10 eth1v10v6 10 MASTER 100 21s
vyos@vyos:~$ show interfaces
Codes: S - State, L - Link, u - Up, D - Down, A - Admin Down
Interface IP Address S/L Description
--------- ---------- --- -----------
[....]
eth1v10v4 192.168.10.60/24 u/u
eth1v10v6 2001:ffff::1/64 u/u
[....]
|
|
|
|
|
|
vrrp: T1972: Ability to set IP address on not vrrp interface
|
|
Ability to set virtual_address on not vrrp-listen interface
Add ability don't track primary vrrp interface "exclude-vrrp-interface"
Add ability to set tracking (state UP/Down) on desired interfaces
For example eth0 is used for vrrp and we want to track another eth1
interface that not belong to any vrrp-group
|
|
conntrack_sync_group option not under 'vrrp' section but part of
high-avalability dictionary
|
|
|
|
Add new feature, high-availability virtual-server
Change XML, python and templates
Move vrrp to root node 'high-availability' as all logic are
handler by root node 'high-availability'
|