| Age | Commit message (Collapse) | Author |
|---|
|
The default `bandwidth` and `ceiling` should calculate values
based on <tag> bandwidth but currently it gets the value from
qos.base `/sys/class/net/{self._interface}/speed`
```
set qos policy shaper SHAPER bandwidth '20mbit'
set qos policy shaper SHAPER default bandwidth '95%'
set qos policy shaper SHAPER default ceiling '100%'
```
It causes wrong calculations for class `default` i.e
950Mbit for bandwidth (expected 95% of bandwidth, 19Mbit)
1Gbit for ceil (expected 100% of bandwidth, 20Mbit)
Gets incorrect values
```
r4# tc class show dev eth1
class htb 1:1 root rate 20Mbit ceil 20Mbit burst 1600b cburst 1600b
class htb 1:a parent 1:1 leaf 8053: prio 0 rate 200Kbit ceil 200Kbit burst 1Mb cburst 1600b
class htb 1:b parent 1:1 leaf 8054: prio 7 rate 950Mbit ceil 1Gbit burst 15200b cburst 1375b
```
Fix this
|
|
Typo (missaligned -/_) in the code causes hw-tc-offload to never be enabled in
the underlaying hardware via ethtool.
|
|
|
|
|
|
T5958: QoS add basic implementation of policy shaper-hfsc
|
|
If we have QoS policy shaper class match `vif` (VLAN) we have to
use `basic match "meta(vlan mask 0xfff eq xxx)` instead of
`action policy`
Actual incorrect TC filter:
tc filter add dev eth1 parent 1: protocol all prio 1 action police rate 100000000 burst 15k flowid 1:64
The correct TC filter after fix:
tc filter add dev eth1 parent 1: protocol all prio 1 basic match "meta(vlan mask 0xfff eq 100)" flowid 1:64
|
|
|
|
It is impossible to detect interface speed for some devices
for exmaple virtio interfaces:
```
vyos@r4:~$ cat /sys/class/net/eth1/speed
-1
```
It causes wrong negative calcultaions like:
- bandwidth: -1000000
- 4% of bandwidth: -40000
tc class replace dev eth1 parent 1: classid 1:1 htb rate -1000000
tc class replace dev eth1 parent 1:1 classid 1:a htb rate -40000
Fix this with checking negative value.
Add default interface speed to 1000 Mbit if we cannot detect the
interface speed, the current default value 10 Mbit is too low
for nowadays
|
|
QoS policy shaper-hfsc was not implemented after rewriting the
traffic-policy to qos policy. We had CLI but it does not use the
correct class. Add a basic implementation of policy shaper-hfsc.
Write the class `TrafficShaperHFS`
|
|
|
|
T5865: Moved ipv6 pools to named ipv6 pools in accel-ppp
|
|
image-tools: T5923: update system_console.py for new GRUB file structure
|
|
Moved ipv6 pools to named ipv6 pools in accel-ppp services
|
|
Add util function to set serial console speed in accordance with revised
GRUB file structure; in keeping with the intentions of the config_mode
script, adjust the GRUB var 'console_speed' to only modify ttyS0.
|
|
* Also migrate `address-range` to `range` tag node for consistency with dhcpv4 server syntax
|
|
This uses a more common pattern froma base class while the original code from
0a1c9bc38 ("T5791: DNS dynamic exclude check for dynamic interfaces PPPoE") is
still retained.
|
|
the lease file (#2796)
|
|
dhcp: T3316: T5787: T5912: Extend scope of DHCP options, bugfixes
|
|
|
|
supported by Kea
|
|
|
|
We have not seen the adoption of the https virtual-host CLI option.
What it did?
* Create multiple webservers each listening on a different IP/port
(but in the same VRF)
* All webservers shared one common document root
* All webservers shared the same SSL certificates
* All webservers could have had individual allow-client configurations
* API could be enabled for a particular virtual-host but was always enabled on
the default host
This configuration tried to provide a full webserver via the CLI but VyOS is a
router and the Webserver is there for an API or to serve files for a local-ui.
Changes
Remove support for virtual-hosts as it's an incomplete and thus mostly useless
"thing". Migrate all allow-client statements to one top-level allow statement.
|
|
Commit ad9bdfc24 ("T5195: add timeout argument to process_named_running()")
added a 2*10 seconds penalty for every interface test (dhcp and dhcpv6).
This leads to long runs of "make test" after an ISO build.
There is no need to wait 10 seconds for a test that checks for a process
not running. The timeout is there to give the process some time to startup.
|
|
image: T5898: fix kernel-level partition rescan
|
|
pki: T5886: add support for ACME protocol (LetsEncrypt)
|
|
Smoketests heavily rely on process_named_running() so in order to "relax"
system constraints during a test we will add a timeout of 10 seconds for
every testcase provided by base_interfaces_test.py
|
|
This fix moves from partprobe to partx to rescan the partition table on an affected disk.
|
|
The "idea" of this PR is to add new CLI nodes under the pki subsystem to
activate ACME for any given certificate.
vyos@vyos# set pki certificate NAME acme
Possible completions:
+ domain-name Domain Name
email Email address to associate with certificate
listen-address Local IPv4 addresses to listen on
rsa-key-size Size of the RSA key (default: 2048)
url Remote URL (default:
https://acme-v02.api.letsencrypt.org/directory)
Users choose if the CLI based custom certificates are used
set pki certificate EXAMPLE acme certificate <base64>
or if it should be generated via ACME.
The ACME server URL defaults to LetsEncrypt but can be changed to their staging
API for testing to not get blacklisted.
set pki certificate EXAMPLE acme url https://acme-staging-v02.api.letsencrypt.org/directory
Certificate retrieval has a certbot --dry-run stage in verify() to see if it
can be generated.
After successful generation, the certificate is stored in under
/config/auth/letsencrypt. Once a certificate is referenced in the CLI (e.g. set
interfaces ethernet eth0 eapol certificate EXAMPLE) we call
vyos.config.get_config_dict() which will (if with_pki=True is set) blend in the
base64 encoded certificate into the JSON data structure normally used when
using a certificate set by the CLI.
Using this "design" does not need any change to any other code referencing the
PKI system, as the base64 encoded certificate is already there.
certbot renewal will call the PKI python script to trigger dependency updates.
|
|
deployments in cgnat. (#2694)
|
|
VyOS has several services relaying on the PKI CLI tree to retrieve certificates.
Consuming services like ethernet, openvpn or ipsec all re-implemented the same
code to retrieve the certificates from the CLI.
This commit extends the signature of get_config_dict() with a new option with_pki
that defaults to false. If this option is set, the PKI CLI tree will be blended
into the resulting dictionary.
|
|
This extends commit 4ee406470 ("configdict: T5837: add support to return added
nodes when calling node_changed()") so no duplicate list elements get returned.
|
|
|
|
|
|
interfaces
Interfaces matching the following regex (ppp|pppoe|sstpc|l2tp|ipoe)[0-9]+ can
not be used as source-interface for e.g. a tunnel.
The main reason is that these are dynamic interfaces which come and go from a
kernel point of view, thus it's not possible to bind an interface to them.
|
|
We will use _ as CLI level divider. The XML definition filename and also
the Python helper should match the CLI node.
Example:
set interfaces ethernet -> interfaces_ethernet.xml.in
set interfaces bond -> interfaces_bond.xml.in
set service dhcp-server -> service_dhcp-server-xml.in
|
|
After deleting a user account and working with a newly added account, we see
that after rebooting in the previously saved configuration, the user is
re-added but it's home directory might have an old UID set on the filesystem.
This is due to the fact that vyos config does not store UIDs. When adding a
user account to the system we now check if the home directory already exists
and adjust the ownership to the new UID.
|
|
The subnet router anycast address is predefined. Its format is as follows:
| n bits | 128-n bits |
+------------------------------------------------+----------------+
| subnet prefix | 00000000000000 |
+------------------------------------------------+----------------+
The "subnet prefix" in an anycast address is the prefix that identifies a
specific link. This anycast address is syntactically the same as a unicast
address for an interface on the link with the interface identifier set to zero.
Packets sent to the Subnet-Router anycast address will be delivered to one
router on the subnet. All routers are required to support the Subnet-Router
anycast addresses for the subnets to which they have interfaces.
The Subnet-Router anycast address is intended to be used for applications where
a node needs to communicate with any one of the set of routers.
Our code as of now returns the subnet router anycast address as the
first_host_address().
|
|
dhcp: T3316: Support hostname, DUID and MAC address in reservation
|
|
Rename chain level defaults log option from `enable-default-log` to
`default-log` for consistency.
|
|
VyOS is a routing (packet pushing) platform, thus supporting EEE which
potentially causes issues is not a good idea. Some recent Intel drivers enable
EEE by default, thus we will disable this for every NIC supporting EEE.
|
|
In the past, node_changed() suggested it would also return nodes that got added
(function comment) but in reality only deleted keys got accounted for.
This commit changes the signature and adds an argument expand_nodes to specify
the users interest of a node was deleted (default), added (expand_nodes=Diff.ADD)
or even both (expand_nodes=Diff.ADD|Diff.DELETE).
|
|
node_changed() will return a list of changed keys under "path". We are not
always interested what changed, sometimes we are only interested if something
changed at all, that what vyos.configdict.is_node_changed() is for.
|
|
Reinstate support for hostname in DHCP reservation. Having `hostname` in
allows for server-side assignment of hostname. This is useful for static
lookup of hostname.
Ensure that hostname is a valid FQDN (doesn't have underscore, etc.)
Additionally, support using either of DUID or MAC address for
reservation. While MAC address is typically used for IPv4, and DUID is
typically used for IPv6, either of them can be used in IPv4 and IPv6
reservations in Kea.
|
|
T5828: fix grub installation on arm64-efi machines
|
|
|
|
|
|
Since the migration of GRUB handling to vyos-1x, the grub install
sequence has hardcoded references to x86.
Change the GRUB sequence so it can work on arm64 as well.
|
|
T5249: Add rollback-soft feature
|
|
* Move Kea socket permission change on-demand and speed up conf scripts
* Fix issue with DHCP reservations when no `ip-address` value
|
|
The current implementation is wrong as it searches `master` in the
iproute2 JSON output. It is a worng as it could include bridges
or bonding interfaces
Add the more strict search `info_slave_kind == vrf`
|
