Upstream merge 2020-02-18 #93
Conversation
When config is read, the config file is parsed first. After these settings are loaded, the CLI settings are parsed and override any settings read from the file. This is fine, except for the CLI settings may be defaulted, which will just override any specified config file settings with defaults. This patch addresses the problem by checking during config override if the the configuration option has a default specified, and if so, ignores overriding the current config option. Additionally, this patch adds real default values for settings. Signed-off-by: Tim Rozet <trozet@redhat.com>
With this change, any config setting is checked against its default to determine if the value is valid when parsed. The only case where empty string values are considered "unset" is with environment variables. Signed-off-by: Tim Rozet <trozet@redhat.com>
This commit implements the clustered database as a statefulsets with
replicas set to 3. The PodManagementPolicy of OrderReady is too
restrictive for how RAFT works, so instead Parallel policy is used.
The existing ovnkube-db service acts as the headless Service for the
statefulset. The pods are scheduled on the nodes with labels
`ovn.org/ovnkube-db=true`.
Implementation notes:
1. ovnkube-db-0, the first pod in the statefulset creates the initial
cluster DB. If an existing standalone DB is present, then it converts it
into the clustered DB.
2. ovnkube-db-1 ... ovnkube-db-{n-1} pods wait until the ovnkube-db-0
has successfully started and then only join the cluster.
3. ovnkube-db-{n-1}, the last pod in the statefulset will ensure that
all of the pods in the statefulet are up and running and then only
publish the ovnkube-db endpoints to the ovnkube-db service.
4. Like before, all of the ovnkube-nodes and ovnkube-master are waiting
on the endpoints to be published. Once the endpoints are published, they
proceed further.
Signed-off-by: Yun Zhou <yunz@nvidia.com>
Signed-off-by: Girish Moodalbail <gmoodalbail@nvidia.com>
It's useful to see successful logs too. Signed-off-by: Dan Williams <dcbw@redhat.com>
Clustered database (aka RAFT) support for OVN DBs
Girish identified a race where the initial object Add events that are sent when a handler is attached to an informer were not properly locked against incoming events to that informer from the apiserver. To fix this send the initial Add events while the informer's lock is held. Signed-off-by: Dan Williams <dcbw@redhat.com>
Previously OVN Kubernetes assumed the node's name in kubernetes was the node's FQDN, which isn't necessarily true, with this change we read it from the kubernetes API, which is more reliable. Signed-off-by: Juan Luis de Sousa-Valadas Castaño <jdesousa@redhat.com>
ci: always export logs
Signed-off-by: Tim Rozet <trozet@redhat.com>
Signed-off-by: Girish Moodalbail <gmoodalbail@nvidia.com>
7358508 changed GatewayReady() from using ovn-nbctl to using ovs-ofctl, but checked for an IPv4-specific rule. Fix it to use IPv6 syntax when checking for IPv6 CIDRs. Signed-off-by: Dan Winship <danw@redhat.com>
We're not ready until the messages channel is closed. Signed-off-by: Dan Winship <danw@redhat.com>
Ensure kind logs are always exported
remove unused methods from KClient
kube-proxy is not necessary for OVN. Signed-off-by: Tim Rozet <trozet@redhat.com>
Address sets were being deleted before the port group. The port group
contained an ACL which referenced the address set. This would cause OVN
controller to issue:
2020-02-17T16:31:09.160Z|00032|lflow|WARN|error parsing match "!ct.new
&& ct.est && !ct.rpl && ct_label.blocked == 0 && (ip4.src ==
{$a2376695646506571724, $a6566950664295833471} && outport ==
@a5584531700272134285)": Syntax error at `$a6566950664295833471'
expecting address set name.
With this patch address set is deleted after the associated ACL in the
port group is removed.
See ovn-org/ovn#35
Signed-off-by: Tim Rozet <trozet@redhat.com>
Remove kube-proxy in KIND deployment
Fix delete order for network policy
…ults Fixes accidental override of file conf settings with defaults
Signed-off-by: Dan Winship <danw@redhat.com>
other-config:subnet takes a CIDR string, but other-config:ipv6_prefix just takes an IP address (and just assumes a /64 prefix). Fix this. (Fixes dynamic assignment of IPv6 addresses.) Also belatedly fix syncNodes which was always looking for "other-config:subnet" even on IPv6. Since other-config:subnet and other-config:ipv6_prefix don't behave the same way and so need separate code (and because they're used less now than when Russell originally wrote that patch anyway), remove config.OtherConfigSubnet(). Signed-off-by: Dan Winship <danw@redhat.com>
add a -no-host-subnet-nodes flag which takes a label and causes ovn-kubernetes to not assign a hostsubnet to that node. This will be used in cases where ovn is not supported and nodes can manage there own subnets Signed-off-by: Jacob Tanenbaum <jtanenba@redhat.com>
add the ability for ovn-kubernetes to ignore a node by label
config: don't let environment affect test results
Signed-off-by: Dan Winship <danw@redhat.com>
fix IPv6 subnet dynamic addressing
The annotator provides a better-encapsulated interface for setting or removing multiple annotations on a node from disconnected functions since it bundles them all up into a single set and/or delete call. Signed-off-by: Dan Williams <dcbw@redhat.com>
Doesn't check for OVS/OVN utilities; to be used for components that don't need those. Signed-off-by: Dan Williams <dcbw@redhat.com>
|
/retest |
dcbw
force-pushed
the
upstream-merge-2020-02-18
branch
from
289a30f
to
af32c8c
Compare
|
Registry returns 500 when pushing the ovnkube image. /retest |
|
We're still getting upstream CI. Didn't we agree to delete/revert that in the merge? |
@danwinship not that I recall but I may have been on PTO at that time. But we can kill it since the openshift e2e tests should be more comprehensive. |
/test e2e-gcp-ovn |
|
/test e2e-aws-ovn |
|
|
|
/retest |
4 similar comments
|
/retest |
|
/retest |
|
/retest |
|
/retest |
|
/test e2e-gcp-ovn |
2 similar comments
|
/test e2e-gcp-ovn |
|
/test e2e-gcp-ovn |
|
@dcbw There are OOMKilled problems that the node team is working on. See slack. Its not likely anything will pass until they are done. |
|
/lgtm |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: danwinship, dcbw The full list of commands accepted by this bot can be found here. The pull request process is described here
Needs approval from an approver in each of these files:
Approvers can indicate their approval by writing |
|
/retest Please review the full test history for this PR and help us cut down flakes. |
|
@dcbw: The following test failed, say
Full PR test history. Your PR dashboard. Please help us cut down on flakes by linking to an open issue when you hit one in your PR. Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. I understand the commands that are listed here. |
@openshift/networking @danwinship