I have IPVS configured as:
$ ipvsadm --save
-A -t localhost:<port> -s rr
-a -t localhost:<port> -r <other_hostname>:<port> -m -w 1
It seems to be running:
$ ipvsadm
IP Virtual Server version 1.2.1 (size=4096)
Prot LocalAddress:Port Scheduler Flags
-> RemoteAddress:Port Forward Weight ActiveConn InActConn
TCP localhost:<port> rr
-> <other_hostname>:<port> Masq 1 1 0
but that 1 'ActiveConn' hangs indefinitely, while a direct request responds quickly and as expected:
$ curl https://<other_hostname>:<port>
<response>
$ curl https://localhost:<port>
[hangs]
Nothing seems wrong in dmesg:
$ dmesg | grep IPVS
[ 4795.060820] IPVS: Registered protocols (TCP, UDP, SCTP, AH, ESP)
[ 4795.060932] IPVS: Connection hash table configured (size=4096, memory=64Kbytes)
[ 4795.061202] IPVS: ipvs loaded.
[ 4795.067663] IPVS: [rr] scheduler registered.
And nothing in journalctl from ipvsadm.
It seems to get stuck in state SYN_RECV:
$ ipvsadm -lnc
IPVS connection entries
pro expire state source virtual destination
TCP 00:56 SYN_RECV 127.0.0.1:58322 127.0.0.1:<port> <other_ip>:<port>
It seems that it isn't a configuration issue with the real server, since it's an issue with all modes (route, masquerade, tunnel) and testing with httpbin.org rather than my own.
What can cause a request to an IPVS to hang or fail when the 'real server'(s) underlying it respond OK, and how can it be debugged?
