VIP still assigned when vip-manager service is stopped

[sdayvz]

Hi,
I use vip-manager 2.8.0 and consul 1.19.0
When I stopped the vip-manager service, the VIP still assigned on server.
I can see the VIP like this:

2: ens224: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc mq state UP group default qlen 1000
    link/ether 00:50:56:b0:0a:a1 brd ff:ff:ff:ff:ff:ff
    altname enp19s0
    inet 10.10.10.10/24 brd 10.88.0.255 scope global noprefixroute ens224
       valid_lft forever preferred_lft forever
    inet 10.10.10.11/24 scope global secondary ens224
       valid_lft forever preferred_lft forever

Here is the vip-manager settings:

interval: 1000
trigger-key: "/patroni/dengage/leader"
trigger-value: "dengage-01"
ip: 10.10.10.11 
netmask: 24 
interface: ens224
hosting-type: basic 
dcs-type: consul
dcs-endpoints:
  - http://127.0.0.1:8500
consul-token: "f62e9800-282a-2baa-5564-71a79f87092e"
retry-num: 2
retry-after: 250  #in milliseconds
verbose: false

Is this a bug? If not, is there a setting to delete VIP automatically?
Thanks.

[pashagolub]

UPD: vip-manager removes VIP as the cleanup routine.

vip-manager doesn't remove VIP because there was no such a condition in consul. vip-manager does only one job: adds vip when it needed to be added, deletes vip when it should be deleted.

In what way stopping the vip-manager affects the patroni cluster?

[sdayvz]

Hi @pashagolub ,
I just wanted to try adding and removing VIP to the server.
If I want to stop using vip-manager, I should delete VIP manually, right?

[pashagolub]

Well, yeah. Or make sure consul is in a state when VIP will be removed automatically

[pashagolub]

I've checked the code and I was wrong. As a cleanup routine vip-manager deconfigures virtual IP:

vip-manager/ipmanager/ip_manager.go

Line 58 in 7ecae13

m.configurer.deconfigureAddress()

Now the question is: why it doesn't in your case? Are there any additional information in logs? Crash maybe?

[balinorLiang]

When vip-manager exits, the deconfigureAddress function is supposed to run the cleanup routine, which executes the command:
ip addr del 192.168.1.101/24 dev eth0
However, this command fails to execute successfully. The log only shows got error: without any further details (no ExitError and output is empty).
Then @sdayvz and I faced with this issue, the VIP still assigned on server when vip-manager exited smoothly.

[gaoxueyu]

I think this is a bug. Although it doesn't happen every time, I frequently encounter VIP unbinding failures in the scenario of gracefully stopping vip-manager—that is, the vip-manager has already stopped, but the VIP still remains. Do we have any plans to fix this issue in the future?

[pashagolub]

We would like to fix all the bugs. Do you have details how can we reproduce it?

[gaoxueyu]

We would like to fix all the bugs. Do you have details how can we reproduce it?

I could reproduce this issue via #326.

First Scenario:
Following are the detailed step:

[Unit]
Description=vip-manager
After=network.target
After=network-online.target
Wants=network-online.target

[Service]
Type=simple
User=root
Group=root
ExecStart=/usr/local/vip-manager/vip-manager --config=/usr/local/vip-manager/vip-manager.yml
Restart=on-failure
KillSignal=SIGINT

[Install]
WantedBy=multi-user.target

systemctl stop vip-manager

second Scenario:
Using kill -9 to smoothly shutdown could also reproduce.

Thanks,
Hope

[gaoxueyu]

Could you please provide a brief update on the status of this bug? If available, I’d also appreciate it if you could share the expected timeline for resolution.

[tpo]

FYI - it should be possible to contact Cybertec and place a commercial order with them to get this resolved if you really require to have this resolved at some specific date and can't do it yourself.