2
0
mirror of https://github.com/xcat2/xcat-core.git synced 2026-11-22 02:12:16 +00:00
Commit Graph

3 Commits

Author SHA1 Message Date
Daniel Hilst 21359734b2 test(xcat-core): cover the dual-stack DHCPv6 boot server on a management node
makedhcp_dhcpv6_boot_intent declares a node with no tftpserver, so it measures only the
<xcatmaster> path. A dual-stack node keeps an IPv4 tftpserver, and that is the path on which
makedhcp exited 1 and dropped the boot URL.

makedhcp_dhcpv6_dual_stack_boot_server declares a grub2 node with an IPv6 address and an IPv4
tftpserver, then asserts that makedhcp exits 0, that Kea accepts the configuration, and that the
bootfile-url names the IPv6 address of the management node facing the node. Both values are
compared exactly, so a dropped option and a wrong authority each fail.

The test addresses move to the documentation ranges at the same time. The three networkutils
tests carried a link-local address read off a lab NIC, and dhcp_kea_plugin_intent.t carried two
addresses from the management network. They are fixtures that nothing dials, so this only keeps
the internal topology out of the upstream tree.

Signed-off-by: Daniel Hilst <392820+dhilst@users.noreply.github.com>
2026-10-27 05:35:23 -03:00
Daniel Hilst 5949f79005 fix(ipv6): the link-local order is applied to one resolver path of two
getipaddr reaches the resolver through Socket6 when that module is present and
through core Socket otherwise, and only the core path ordered its results. A
management node with Socket6 installed therefore still answered
fe80::4054:ff:fec9:9ad2%mgmt0 for its own name, which is what xcatconfig wrote
into site.master. Measured on an AlmaLinux 10.2 node with perl-Socket6: the
previous commit was installed and getipaddr returned the link-local address
unchanged.

_socket6_addrinfo_routable_first reorders the flat five-element records
Socket6::getaddrinfo answers, by the same rule the core path uses. A list whose
length is not a whole number of records is returned untouched.

networkutils_linklocal_preference.t covers the record reorder with a stubbed
Socket6::getnameinfo. Making the link-local test constant false fails 2 of its
assertions. The call site itself has no unit test; the cluster run is what
shows site.master routable.

Signed-off-by: Daniel Hilst <392820+dhilst@users.noreply.github.com>
2026-10-26 15:27:19 -03:00
Daniel Hilst 1253af09c7 fix(ipv6): site.master takes the link-local address of the management node
On an EL10 management node whose own name resolves to both families,
xcatconfig writes a link-local IPv6 address into site.master and
site.nameservers. Measured on an AlmaLinux 10.2 node: getent reports
fe80::4054:ff:fec9:9ad2%2 before 192.168.201.154, and
xCAT::NetworkUtils->getipaddr($hostname) answers
fe80::4054:ff:fec9:9ad2%mgmt0, which xcatconfig stores. A node on another link
cannot reach that address, and the %zone form is not a literal isIpv6addr or
any consumer of site.master accepts.

getipaddr resolves through Socket::getaddrinfo with AF_UNSPEC when no family is
asked for, and returned the first entry getaddrinfo gave. prefer_routable_addresses
orders address literals so a routable one comes before a link-local one and
removes the zone identifier, and _addrinfo_routable_first applies the same order
to the getaddrinfo entries before getipaddr reads the first. A link-local
address is still returned when the host has no other.

networkutils_linklocal_preference.t covers the order, the zone identifier and
both link-local ranges. Removing the routable split fails 4 of its 10
assertions, removing the zone strip fails 3, and treating 169.254.0.0/16 as
routable fails 1.

Signed-off-by: Daniel Hilst <392820+dhilst@users.noreply.github.com>
2026-10-26 15:27:19 -03:00