The outstanding kernel panic should be already fixed in recent enough
kernels by [0]. To make the test safe to run anywhere, let's implement
a simple kernel version check and run the test only if we're running
with at least kernel 6.x. The patch might be in some 5.x kernels as
well, but let's be on the safe side and use 6.x as a baseline here
(which is currently the case for Arch and Fedora Rawhide anyway).
[0] https://lore.kernel.org/netdev/7b3fd03e1a46047e5ffe2a389fe74501f0a93206.1656519221.git.sd@queasysnail.net/T/#u
(s) is just ugly with a vibe of DOS. In most cases just using the normal plural
form is more natural and gramatically correct.
There are some log_debug() statements left, and texts in foreign licenses or
headers. Those are not touched on purpose.
That's not necessary. Moreover, if the socket units are stopped in
`setUpModule()`, then there exists a short timespan that we cannot call
`udevadm control`, as the control socket may not be opened yet.
If we run whole tests, then the first test is
NetworkctlTests.test_altname, and it calls `udevadm control` in `setUp()`.
Hence, the test may fail.
Fixes https://github.com/systemd/systemd-centos-ci/pull/512#issuecomment-1191591008.
Several DHCP client tests change the system timezone.
Let's save the current timezone at the beginning, and restore it with
the saved value at the end.
- introduce several helper functions
- do not list unit files, but remove the runtime unit directory in
tearDown().
- do not list used interfaces, but remove all interfaces previously not
exists in tearDown().
- save routes and routing policy rules before running tests, and flush
unnecessary routes and rules in each tearDown() calls.
- drop many time.sleep() calls.
- call tearDown() after each sub tests.
- shorten code.
- several coding style fixes.
- etc, etc...
Hopefully, this improves performance of the test.
`global` is needed only when assigning a new value to the global
variable; it's not necessary when modifying a mutable object (in our
case we just append items to the global list).
Otherwise, this easily trigger another exception:
```
======================================================================
ERROR: test_erspan_tunnel_v0 (__main__.NetworkdNetDevTests)
----------------------------------------------------------------------
Traceback (most recent call last):
File "./test/test-network/systemd-networkd-tests.py", line 686, in wait_online
check_output(*args, env=env)
File "./test/test-network/systemd-networkd-tests.py", line 65, in check_output
return subprocess.check_output(command, universal_newlines=True, **kwargs).rstrip()
File "/usr/lib64/python3.6/subprocess.py", line 356, in check_output
**kwargs).stdout
File "/usr/lib64/python3.6/subprocess.py", line 438, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['/usr/lib/systemd/systemd-networkd-wait-online', '--timeout=20s', '--interface=erspan99:routable', '--interface=erspan98:routable', '--interface=dummy98:degraded']' returned non-zero exit status 1.
During handling of the above exception, another exception occurred:
Traceback (most recent call last):
File "./test/test-network/systemd-networkd-tests.py", line 1808, in test_erspan_tunnel_v0
self.wait_online(['erspan99:routable', 'erspan98:routable', 'dummy98:degraded'])
File "./test/test-network/systemd-networkd-tests.py", line 689, in wait_online
output = check_output(*networkctl_cmd, '-n', '0', 'status', link.split(':')[0], env=env)
File "./test/test-network/systemd-networkd-tests.py", line 65, in check_output
return subprocess.check_output(command, universal_newlines=True, **kwargs).rstrip()
File "/usr/lib64/python3.6/subprocess.py", line 356, in check_output
**kwargs).stdout
File "/usr/lib64/python3.6/subprocess.py", line 438, in run
output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['/usr/bin/networkctl', '-n', '0', 'status', 'erspan99']' returned non-zero exit status 1.
```
Not sure when the issue was fixed.
- kernel-3.10 on CentOS 7 has the issue,
- kernel-4.18 on CentOS 8 works fine.
Note, the workaround dropped by the commit is not incomplete:
with an old kernel which has the issue, all non-prefix routes are
configured on the specified route table, but the prefix route is
configured on the main table. That should not work for most cases,
hence, the workaround is mostly meaningless.
There is no reason networkd refuses that. Especially, when multiple
downstream interfaces are connected to the same network, it is natural to
assign the same subnet prefix to them.
Prompted by #22571.
If we assign our own test runner, passing arguments stops working
as unittest won't instantiate its own test runner after it parses
the arguments from sys.argv.
Consequence is that the tests will write to stderr now instead of
stdout since it doesn't seem possible to configure the stream that
unittest.main() will instantiate its test runner with so it'll
default to sys.stderr.
Add the "Isolated" parameter in the *.network file, e.g.,
[Bridge]
Isolated=true|false
When the Isolated parameter is true, traffic coming out of this port
will only be forward to other ports whose Isolated parameter is false.
When Isolated is not specified, the port uses the kernel default
setting (false).
The "Isolated" parameter was introduced in Linux 4.19.
See man bridge(8) for more details.
But even though the kernel and bridge/iproute2 recognize the "Isolated"
parameter, systemd-networkd did not have a way to set it.
The warning is correct, since we don't inherit the necessary
unittest.TestCase class, but that's on purpose, since the Utilities
class is not supposed to be instantiated on its own, but should
complement other classes' definitions which do inherit from the
unittest.TestCase class.
It currently works because `\(` and `\)` are not valid escape sequences,
so they're not treated differently. Using raw strings (or double
backslashes) is a more correct solution.
Multiline comments are converted to docstrings only when they're the
first statement in a function/method. Even though they're still a no-op
otherwise, let's use "true" comments to make pylint happy.
If wait_operstate() is called super quickly after ip command, then the
up/down state may not be changed and propagated to networkd, and
wait_operstate() mistakenly pass with the previous state.
To avoid such situation, wait for a while to make networkd actually
detect the interface brought up/down.
This fixes the following race:
1. when a dummy interface is created, it is initially down state,
2. hence, wait_operstate() may pass before the link is activated,
3. and the ip command bring up the interface before the activation,
4. and networkd activates, that is, brings down the interface,
5. thus, next wait_operstate() timedout, as it waits for the interface up.
To fix the race, let's wait the link is activated, before enter the loop
of wait_operstate().
Fixes#22267.
Strictly speaking, this breaks the backward compatibility, but I guess
in most cases people already sets Scope=link for such routes.
This behavior matches with how 'ip route' command adds such route by
default.
Prompted by https://twitter.com/jplitza/status/1480500562391179270.
People often assigns the MAC address of the enslaved interface to e.g.
bridge interface. So, the local assignment bit should not be adjusted.
Fixes#21649.
Depending on the location of the original build dir, either ProtectHome=
or ProtectSystem= may get in the way when creating the gcov metadata
files.
Follow-up to:
* 02d7e73013
* 6c9efba677
With `ProtectSystem=strict` gcov is unable to write the *.gcda files
with collected coverage. Let's add a yet another switch to make such
restriction less strict to make gcov happy.
This addresses following errors:
```
...
systemd-networkd[272469]: profiling:/systemd-meson-build/src/shared/libsystemd-shared-249.a.p/binfmt-util.c.gcda:Cannot open
systemd-networkd[272469]: profiling:/systemd-meson-build/src/shared/libsystemd-shared-249.a.p/base-filesystem.c.gcda:Cannot open
systemd-networkd[272469]: profiling:/systemd-meson-build/src/shared/libsystemd-shared-249.a.p/barrier.c.gcda:Cannot open
systemd-networkd[272469]: profiling:/systemd-meson-build/src/shared/libsystemd-shared-249.a.p/ask-password-api.c.gcda:Cannot open
systemd-networkd[272469]: profiling:/systemd-meson-build/src/shared/libsystemd-shared-249.a.p/apparmor-util.c.gcda:Cannot open
systemd-networkd[272469]: profiling:/systemd-meson-build/src/shared/libsystemd-shared-249.a.p/acpi-fpdt.c.gcda:Cannot open
...
```
When playing around with the coverage-enabled build I kept hitting
an issue where dnsmasq failed to start because the previous instance was
still shutting down. This should, hopefully, help to mitigate that.
I want to mark some files to be ignored for licensing purposes,
e.g. output from fuzzers and other samples. By using the gitattribute
machinery for this we don't need to design a custom protocol:
$ git check-attr generated test/test-sysusers/unhappy-*
test/test-sysusers/unhappy-1.expected-err: generated: set
test/test-sysusers/unhappy-1.input: generated: unspecified
test/test-sysusers/unhappy-2.expected-err: generated: set
test/test-sysusers/unhappy-2.input: generated: unspecified
test/test-sysusers/unhappy-3.expected-err: generated: set
test/test-sysusers/unhappy-3.input: generated: unspecified
Those are all consumed by our parser, so they all support comments.
I was considering whether they should have a license header at all,
but in the end I decided to add it because those files are often created
by copying parts of real unit files. And if the real ones have a license,
then those might as well. It's easier to add it than to make an exception.
Previously, when Priority= is unspecified, networkd configured the rule with
the highest (=0) priority. This commit makes networkd distinguish the case
the setting is unspecified and one explicitly specified as Priority=0.
Note.
1) If the priority is unspecified on configure, then kernel dynamically picks
a priority for the rule.
2) The new behavior is consistent with 'ip rule' command.
Replaces #15606.
There are nothing we can configure in udevd for loopback interfaces;
no ethertool configs can be applied, MAC address, interface name should
not be touched.
Now, RoutesToDNS= and RoutesToNTP= are enabled by default on DHCPv4
client. So, if DHCP server picks up DNS or NTP servers from uplink,
then the routes may break CI environment.
Hopefully fixes#19463.
As LLDP thing does not get involved in the link status, `networkctl lldp`
may not provide an expected information even if the link is in
'configured' state.
Fixes#17360.
Previous commits changed the dhcpv4 retransmission algorithm to be
slightly slower, changing the amount of time it takes to notify
systemd-networkd that the dhcpv4 configuration has (transiently)
failed from around 14 second up to 28 seconds.
Since the test_dhcp_client_with_ipv4ll_without_dhcp_server test
configures an interface to use dhcpv4 without any operating dhcpv4
server running, it must increase the amount of time it waits for
the test interface to reach degraded state.
As interfaces will be reconfigured asynchronously after `networkctl reload`.
So, right after `networkctl reload` is finished, interfaces may be still
in 'configured' state with the old .network files.
They are not really boolean, because we have both ipv4 and ipv6, but
for each protocol we have either unset, no, and yes.
From https://github.com/systemd/systemd/issues/13316#issuecomment-582906817:
LinkLocalAddressing must be a boolean option, at least for ipv4:
- LinkLocalAddressing=no => no LL at all.
- LinkLocalAddressing=yes + Static Address => invalid configuration, warn and
interpret as LinkLocalAddressing=no, no LL at all.
(we check that during parsing and reject)
- LinkLocalAddressing=yes + DHCP => LL process should be subordinated to the
DHCP one, an LL address must be acquired at start or after a short N
unsuccessful DHCP attemps, and must not stop DHCP to keeping trying. When a
DHCP address is acquired, drop the LL address. If the DHCP address is lost,
re-adquire a new LL address.
(next patch will move in this direction)
- LinkLocalAddressing=fallback has no reason to exist, because LL address must
always be allocated as a fallback option when using DHCP. Having both DHCP
and LL address at the same time is an RFC violation, so
LinkLocalAdressing=yes correctly implemented is already the "fallback"
behavior. The fallback option must be deprecated and if present in older
configs must be interpreted as LinkLocalAddressing=yes.
(removed)
- And for IPv6, the LinkLocalAddress option has any sense at all? IPv6-LL
address aren't required to be always set for every IPv6 enabled interface (in
this case, coexisting with static or dynamic address if any)? Shouldn't be
always =yes?
(good question)
This effectively reverts 29e81083bd. There is no
special "fallback" mode now, so the check doesn't make sense anymore.
The link state file is always removed when networkd is stopping. So,
the deserialization logic does not work. Moreover, the ADDRESSES=
entry is not used by sd-network, so serialization is also not necessary.
Before this commit, event when Gateway=_dhcp4 or _ra is set, the
route was configured with 'protocol static', and other properties
specified by RouteTable=, RouteMTU=, or etc, were ignored.
This commit makes set the route protocol based on the protocol the
gateway address is obtained, and apply other settings if it is not
explicitly specified in the [Route] section.
