Skip to content

Allow OSPF or OSPFv3 adjacency to form#1034

Open
birdofprey wants to merge 150 commits intomainfrom
stable-0.8
Open

Allow OSPF or OSPFv3 adjacency to form#1034
birdofprey wants to merge 150 commits intomainfrom
stable-0.8

Conversation

@birdofprey
Copy link
Copy Markdown

@birdofprey birdofprey commented Oct 21, 2022

vi /etc/firewalld/service/ospf.xml

<?xml version="1.0" encoding="utf-8"?>
<service>
  <short>OSPF</short>
  <description>Allow OSPF or OSPFv3 adjacency to form</description>
  <protocol value="ospf"/>
  <destination ipv4="224.0.0.5"/>
  <destination ipv4="224.0.0.6"/>
  <destination ipv6="ff02::5"/>
  <destination ipv6="ff02::6"/>
</service>

erig0 added 30 commits November 5, 2019 08:43
Fixes: 0c07b70 ("feat: CLI: add "helper" support for services")
Fixes: rhbz 1769520
(cherry picked from commit 6bfffe6)
Use the change interface call instead of add interface. This lets NM
override the permanent interface assignment.

Fixes: rhbz 1773809
(cherry picked from commit a3265da)
If nft fib is not available the test was checking for "system", but it
always yields "no".

Fixes: 6cd756b ("chore: deprecate AutomaticHelpers")
(cherry picked from commit 58c19a0)
Calling modprobe is problematic inside a container. Just check if the
running kernel is >4.18 as this is when NAT coexistence was fixed.

(cherry picked from commit 2b7d150)
Match more specifically so we don't accidentally match a debug log that
also has "ERROR:" or "WARNING:" as is common for modprobes.

(cherry picked from commit 5f67a78)
This is just a dummy target at the moment.

(cherry picked from commit df13ebc)
Introduce "cleanup_late" for high level stuff not used by test cases.

(cherry picked from commit ebe4ee5)
Useful for NetworkManager integration tests.

(cherry picked from commit 608f007)
These use the target "check-integration". We use a separate target
because these tests may be destructive to the host. The plan is to run
them from within the "check-container" target.

(cherry picked from commit ab6c22b)
Coverage for rhbz 1773809

(cherry picked from commit ea97fb2)
This ties the integration tests into the "check-container" target.

NOTE: We force "-j1" because the integration tests must be run serially.
(cherry picked from commit c1c8156)
We're now using Weblate.

(cherry picked from commit a43d77f)
We need to make sure the pre-requisite chains exist before generating
the rule for the mark action.

Fixes: #567
(cherry picked from commit 4997385)
(cherry picked from commit cdf3227)
erig0 and others added 30 commits August 6, 2020 09:13
Fixes: 3a0e79b ("fix: core: rich: Catch ValueError on non-numeric priority values")
(cherry picked from commit e1562ba)
newer networkmanager ships with this zone. as such, if nm is installed
the expected "get zones" output changes.

(cherry picked from commit a609c15)
pip changed behavior and doesn't allow multiple eggs from the same file.
Don't use pip and install directly from setup.py.

(cherry picked from commit 8d0f6c1)
Fixes: 6e279ef ("test(regression/rhbz1541077): correctly use macros")
Fixes: dddba7b ("fix(cli): add ipset type hash:mac is incompatible with the family parameter")
(cherry picked from commit fae16b5)
Fixes: 87ec14d ("test(rich): icmptypes with one family")
(cherry picked from commit a47819d)
Otherwise we may get runtime errors because the running kernel doesn't
support the ict. Use the permanent ict definition so we allow the case
where ip6tables is missing or not available. Explicit usage of an ict
not supported by the kernel will still fail to apply at runtime
(iptables complains), but if ip6tables is missing we don't attempt to
apply the ipv6 rules thus avoiding the issue.

(cherry picked from commit fdc4480)
One scenario is if IPv6 is not available, but we specify an icmp-type
that is ipv6 only, then we'll still attempt to call the IPv6 backend. We
should not do that.

(cherry picked from commit 4fcb27b)
This is a stable only fix. It does not occur on master.
There are various areas that we use list comprehensions to convert
Rich_Rule to rule_str. This isn't cheap. Let's just cache the rule_str
and avoid the cost.

Fixes: rhbz 1871298
(cherry picked from commit 5402724)
Coverage for rhbz 1871298.
Verify we can parse a large amount of rich rules in a reasonable time.

This test took 3m before the fix and now takes 18s after the fix.
Considering it "failed" after 45s should give us plenty of headroom.

(cherry picked from commit ece3097)
We were not filling the runtime ipv6 icmptypes list if the active
backend was nftables and ip6tables wasn't available. This caused "ipv6"
to be dropped from the supported ipvs/destinations for the icmptype.
This also caused runtimeToPermanent to fail because the runtime
icmptypes definition dropped "ipv6" causing runtimeToPermanent to copy
the runtime icmptype to permanent because they were different... this
caused sanity checks on the permanent configuration to fail.

(cherry picked from commit c92d43d)
This is to uncover bugs in which we're using nftables backend but
mistakenly checking ip6tables for information and ip6tables is not
available.

(cherry picked from commit 44f4c4c)
Reported-by: D. Hugh Redelmeier <[email protected]>
(cherry picked from commit a7b12b8)
Replace invalid method names for both 'queryEntry' and 'queryIPSet'.

Fixes: #693
(cherry picked from commit 6fc82d2)
Fixes: rhbz 1596304
Fixes: #480

(cherry picked from commit ac59608)
We need kernel >=5.3 for NAT in the "inet" family. At the same time
we're out of travis CI minutes. Move completely to github actions.

(cherry picked from commit c313bc2)
nftables supports matching the destination MAC, but iptables does not.
As such, lift the restriction from nftables. For iptables, gracefully
ignore the scenarios in which we attempt to match destination MAC.

Fixes: #703
Fixes: df4aefc ("improvement(ipXtables): add utility function match sources")
Fixes: 1582c5d ("feat: nftables: convert to libnftables JSON interface")

Co-authored-by: Eric Garver <[email protected]>
(cherry picked from commit 20151fb)
Generating the table was taking an unreasonable amount of memory.
Stripping C0 and C1 should cover most scenarios while limiting memory
usage.

Fixes: ac59608 ("fix(rich): non-printable characters removed from rich rules")
(cherry picked from commit 015704b)
Where real interface means linux interface capable of having an IP
address and does not exceed IFNAMSIZ.

Fixes: rhbz 1928860
(cherry picked from commit f18f1cc)
Coverage: rhbz 1928860
(cherry picked from commit 7566d3d)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.