incorporate maghemite #533 #9449
Conversation
| &self, | ||
| opctx: &OpContext, | ||
| sel: ¶ms::BgpRouteSelector, | ||
| _sel: ¶ms::BgpRouteSelector, |
There was a problem hiding this comment.
If we're no longer querying imported routes by ASN, we should propagate that up through the omicron API function networking_bgp_imported_routes_ipv4 so users of the API don't have to specify an ASN that's no longer used for anything. Then we can remove the sel parameter.
There was a problem hiding this comment.
Agreed. However, the change I'm envisioning would be to deprecate the old endpoint (/v1/system/networking/bgp-routes-ipv4) and introduce a new endpoint that will not be specific to BGP or IP version (e.g. /v1/system/networking/routes). I'm not sure exactly what this will look like yet from a Nexus API standpoint, so I put this in as a placeholder for the time being.
Signed-off-by: Trey Aspelund <trey@oxidecomputer.com>
maghemite #533 adds impl PartialEq<rdb_types::Prefix> for oxnet::IpNet, which caused some preexisting type inferences to no longer be resolved unambiguously (now there are multiple PartialEq<T> for oxnet::IpNet). This updates these spots to be explicit about the types being compared. Signed-off-by: Trey Aspelund <trey@oxidecomputer.com>
This PR connects IPv6 routes from the switch port settings API to maghemite through the sync switch configuration RPW.
rcgoodfellow
force-pushed
the
trey/ipv6
branch
from
767a2fc to
ae3aec5
Compare
| let mut switch_zone_addrs = HashMap::new(); | ||
|
|
||
| for addr in switch_zone_addresses { | ||
| let port = match resolver |
There was a problem hiding this comment.
Special scrutiny is warranted here. I think this should be ok as we fallback to the previous constant port when DNS does not work out.
There was a problem hiding this comment.
FWIW, this function will probably go away entirely as we continue to chop away at #8999. In the branch I'm building off of this PR for multi-rack omicron-dev, building the mgd clients will no longer rely on this and will instead look like dpd_clients() in this same file. I think LLDP is the only one left after that, which I also plan to address once MGD is working.
| let mut clients: Vec<(SwitchLocation, mg_admin_client::Client)> = vec![]; | ||
| for (location, addr) in &mappings { | ||
| let port = MGD_PORT; | ||
| let port = match resolver.lookup_all_socket_v6(ServiceName::Mgd).await { |
There was a problem hiding this comment.
Special scrutiny is warranted here. I think this should be ok as we fallback to the previous constant port when DNS does not work out.
There was a problem hiding this comment.
Same as noted above
internet-diglett
left a comment
internet-diglett
left a comment
There was a problem hiding this comment.
Some very minor comments, LGTM
- Connects IPv6 routes from the switch port settings API to maghemite through the sync switch configuration RPW - Adds tests to verify that v6 routes make it all the way to mgd - Fixes some hardcoded ports for mgs and mgd to allow e2e maghemite testing - Fixes `tools/update_maghemite.sh` as maghemite no longer ships openapi definitions as buildomat artifacts. --------- Signed-off-by: Trey Aspelund <trey@oxidecomputer.com> Co-authored-by: Ryan Goodfellow <ryan.goodfellow@oxide.computer>
tools/update_maghemite.shas maghemite no longer ships openapi definitions as buildomat artifacts.