I see something I should clarify directly:
" I see packets going out and replies coming in - but Linux does not appear
to get any traffic (RX counters not increasing)" - when I say "I see
packets going out and replies coming in" that's on the switch side, not in
the zLinux LPAR.
On Tue, Jan 28, 2020 at 9:40 PM Christian Svensson wrote:
> Hi,
>
> I'm operating a z114 with OSA Express 4S 1G and 10G cards.
> I'm having the following issue on both zLinux (kernel 5.3.0) and z/VM 630.
> When TCPIP is restarted, either in VM as "FORCE TCPIP" + "XAUTOLOG TCPIP"
> (or linux as "ip link set down" followed up "up"), or a re-IPL in either
> OS, the IPv6 address will not be pingable again (and no other IPv6
> traffic either).
>
> The things I've noted that will make the instance reply to ping again are:
>
>1. Cycling the port in the switch ("shutdown" + "no shutdown")*
>2. In the VM case, pinging out through the gateway
> 1. This does *not* help in zLinux oddly enough, I see packets going
> out and replies coming in - but Linux does not appear to get any traffic
> (RX counters not increasing)
> 2. Using *another* IPv6 address on zLinux works without issue
>
> * However, it seems like the qeth driver in zLinux is made to forget the
> IP address in case of a link-flap (which is not how conventional network
> cards behave in Linux).
>
> I am pretty confident that it is not the switch that is acting up, I can
> see the traffic counters increasing on the switch side.
> The network is IPv6 only so trying with IPv4 traffic will be quite
> troublesome, and replacing the switch is not really feasible as well.
> First I thought it was some odd issue in my setup of zLinux, but when I
> started testing z/VM and hit this similar issue I started to wonder if this
> is some known issue.
>
> Has anyone seen similar behaviour in their environments?
>
> Thanks,
>
--
For IBM-MAIN subscribe / signoff / archive access instructions,
send email to lists...@listserv.ua.edu with the message: INFO IBM-MAIN
