The hardware nodes are running Debian 7 with OpenVZ 6, CTs are mixed Debian
7/8 and CentOS 6/7.
On Tue, Jan 16, 2018 at 4:32 PM, Narcis Garcia <informat...@actiu.net>
wrote:
> "Debian OpenVZ 6 machines" means the hardware (host) nodes or VPS?
>
>
> El 16/01/18 a les 15:37, Baptiste Ashka ha escrit:
> > Hi,
> >
> > I am encountering a network-related issue with some Debian OpenVZ 6
> > (2.6.32-openvz-042stab127.2-amd64) machines that occurs at seemingly
> > random and that only happens on OpenVZ nodes: all connect() syscalls
> > that try to connect to 127.0.0.1 (IPv4) will timeout. Connecting to ::
> > (IPv6) works, any other 127.x.x.x addresses work. Rebooting the hardware
> > node is the most consistent workaround we have (but it does not fix the
> > issue every time either), stopping a bunch of CTs sometimes also fixes
> > the issue, sometimes weird things happen (e.g. suspend a specific CT,
> > 127.0.0.1 works again, resume the CT, 127.0.0.1 stops working, suspend
> > this CT again, 127.0.0.1 does not work). This does not appear to be a
> > socket count limit issue, from what I can tell.
> >
> > This happens sporadically but consistently on all our machines. Is
> > anyone having this issue/something close to it?
> >
> > strace of a connect() syscall using curl:
> > socket(PF_INET, SOCK_STREAM, IPPROTO_TCP) = 3
> > setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0
> > connect(3, {sa_family=AF_INET, sin_port=htons(8080),
> > sin_addr=inet_addr("127.0.0.1")}, 16) = -1 ETIMEDOUT (Connection timed
> out)
> >
> > Thank you
> >
> >
> > _______________________________________________
> > Users mailing list
> > Users@openvz.org
> > https://lists.openvz.org/mailman/listinfo/users
> >
> _______________________________________________
> Users mailing list
> Users@openvz.org
> https://lists.openvz.org/mailman/listinfo/users
>
_______________________________________________
Users mailing list
Users@openvz.org
https://lists.openvz.org/mailman/listinfo/users
