PING health-monitor in LBaaS Haproxy sends a TCP request to members

[Varun Lodaya]

There are different health-monitors in LBaaS:
1) PING 2) TCP 3) HTTP and 4) HTTPS

I was trying PING health-monitor with Haproxy, but it seem to be sending TCP requests to the members instead of PING requests.

varunlodaya@ubuntu:~$ neutron lb-healthmonitor-show fb5d0e4b-5763-4f38-bf2c-09f9f7ab2e49
+----------------+-----------------------------------------------------------------------------------------------------+
| Field | Value |
+----------------+-----------------------------------------------------------------------------------------------------+
| admin_state_up | True |
| delay | 30 |
| id | fb5d0e4b-5763-4f38-bf2c-09f9f7ab2e49 |
| max_retries | 2 |
| pools | {"status": "ACTIVE", "status_description": null, "pool_id": "93d03b4e-05a5-4691-b36f-1416b96c3751"} |
| tenant_id | 6d560cf5767d4f17a61ebc11c14bc1cc |
| timeout | 5 |
| type | PING |
+----------------+-----------------------------------------------------------------------------------------------------+

The Haproxy config it generates for backend is:

*********
backend 93d03b4e-05a5-4691-b36f-1416b96c3751
        mode http
        balance roundrobin
        option forwardfor
        timeout check 5s
        server 07285f2e-9e9f-41ad-a343-2f1d7296f2d9 10.0.0.4:80 weight 1 check inter 30s fall 2

On the member, I opened tcpdump to check whats being received:

sudo tcpdump -i eth0 -v icmp
sudo: unable to resolve host ubuntu-vm
tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 65535 bytes
^C
0 packets captured
0 packets received by filter
0 packets dropped by kernel

ubuntu@ubuntu-vm:~$ sudo tcpdump -i eth0 -n src host 10.0.0.5 -vv
sudo: unable to resolve host ubuntu-vm
tcpdump: listening on eth0, link-type EN10MB (Ethernet), capture size 65535 bytes
23:03:53.434592 IP (tos 0x0, ttl 64, id 4614, offset 0, flags [DF], proto TCP (6), length 60)
    10.0.0.5.55697 > 10.0.0.4.80: Flags [S], cksum 0xc228 (correct), seq 3491668946, win 29200, options [mss 1460,sackOK,TS val 1552852 ecr 0,nop,wscale 7], length 0
23:03:58.441968 ARP, Ethernet (len 6), IPv4 (len 4), Request who-has 10.0.0.4 tell 10.0.0.5, length 28
23:04:23.439097 IP (tos 0x0, ttl 64, id 30112, offset 0, flags [DF], proto TCP (6), length 60)
    10.0.0.5.55704 > 10.0.0.4.80: Flags [S], cksum 0x3862 (correct), seq 635615873, win 29200, options [mss 1460,sackOK,TS val 1560353 ecr 0,nop,wscale 7], length 0
^C
3 packets captured
3 packets received by filter
0 packets dropped by kernel

[Armando Migliaccio (armando-migliaccio)]

No code, and no activity...can't see this land in RC1

[Raseel (raseel)]

I don't think HAProxy supports "PING" health checks. I could verify that HAProxy supports TCP , HTTP and HTTPS health monitoring. Nowhere was there mention of monitoring using ICMP.
That option must be for one of the 3rd party drivers like Radware or A10 or some such.

[venkata anil (anil-venkata)]

Agree with Raseel. HAproxy doesn't support ICMP based health monitoring.
So modifying the bug as invalid.