TCP connections over IPv6 are not stable
Issue
After update of host packages (yum update), we have troubles sending status to SIP proxy in VMs. This did not happen on older kernel.
The communication diagram is as follows:
- HOST 1 hosts VM1.
- VM1 communicates using eth1/eth1.5.
- This interface is a bridge to brVOIP46 interface on HOST 1.
- brVoIP46 is a bond over eth3 and eth4.
HOST1 GUEST (VM1)
eth3/4 -> bondVOIP46 -> brVOIP46 | ------------- | eth1
eth1.5 (VLAN)
eth1.6 (VLAN)
-
HOST 2 hosts VM2.
- VM2 communicates using eth1/eth1.5.
- This interface is a bridge to brVOIP46 interface on HOST 2.
- brVoIP46 is a bond over eth4 and eth5.
-
HOST 1 eth3,eth4 and HOST 2 eth4,eth5 are connected to a pair of L2/L3 switch.
Scenario 1
Host kernel:
kernel-2.6.32-431.el6.x86_64
VM kernel:
kernel-2.6.32-431.11.2.el6.x86_64
sip_proxy_mon works fine and heartbeat packages are received by VMs.
Scenario 2
After "yum update" on Host:
Host kernel:
kernel-2.6.32-431.11.2.el6.x86_64
VM kernel:
kernel-2.6.32-431.11.2.el6.x86_64
Communication from with heartheat from VM 1 to VM 2 does not work and VM 2 claims VM 1 application "DEAD":
TCP connections over IPv6 from a VM 1 running on a host with new kernel to another VM 2 running with old kernel are not stable, a connection is established (SYN, ACK/SYN, ACK) but data packets are retransmitted due to missing an acknowledgement, it seems data packets are dropped due to a wrong TCP checksum. Connections in an opposite direction are OK.
Environment
- Red Hat Enterprise Linux 6.5
kernel-2.6.32-431.11.2.el6and newer
Subscriber exclusive content
A Red Hat subscription provides unlimited access to our knowledgebase of over 48,000 articles and solutions.
Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.
