Age | Commit message (Collapse) | Author | Files | Lines |
|
|
|
Fix rsync to only retry reaching clusters until the
timeout occurs.
Issue-ID: MULTICLOUD-1226
Signed-off-by: Eric Multanen <eric.w.multanen@intel.com>
Change-Id: I5e8202358cfd8cdd15c9feb4fcbefe2683b7139a
|
|
In DCM.
Essentially refactored two different areas:
- rename apply/terminate functions in module to clarify what they do
- split gets from getAlls in API code of the 5 DCM resource types
And cleaned up here and there.
Issue-ID: MULTICLOUD-1143
Change-Id: I9b72c77ba34a1febd5df4a339e87968ddc4a7891
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
The previous commits were merged in the wrong order, which caused a bad
3-way merge resulting in the arg mismatch as witnessed below:
k8s/src/dcm/pkg/modulepkg/module/cluster.go:263:57: not enough arguments
in call to lcClient.GetLogicalCloudContext
have (string)
want (string, string)
Makefile:13: recipe for target 'all' failed
make: *** [all] Error 2
The CI merge job doesn't run tests before merging, so this slipped in.
Issue-ID: MULTICLOUD-1143
Change-Id: I0168345af6bac7886cba32d958f241687ab24a81
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
|
|
|
|
|
|
Update ResourceBundleState CRD and monitor-deploy.yaml with
csrStatuses.
This was missing from the previous patch implementing CSR support
in Monitor. Depending on how the environment was brought up, attempting
to read the CSR would result in forbidden access.
Issue-ID: MULTICLOUD-1143
Change-Id: I86ffb8df8faa045e57fd553e5784d308e1900ee4
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
|
|
Issue-ID: MULTICLOUD-1143
Change-Id: Iaca9bd6cb3614e0ae825d9231476524ce09db550
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
Provide information about instance resources and Pods inside status
response.
Issue-ID: MULTICLOUD-1177
Signed-off-by: Konrad Bańka <k.banka@samsung.com>
Change-Id: Iee6fd56120d091dddfa6b6d0e4aa7eb36d40e888
|
|
DCM apply/terminate code was still using a hardcoded "test-project"
string for the project name that logical clouds should associate to.
This simple patch makes DCM use the project name provided by the API.
It also checks that the project exists before creating a Logical Cloud.
Issue-ID: MULTICLOUD-1143
Change-Id: I05c8a2309ed07a7c96da30bf2461c43abaccc9d5
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
The /kubeconfig API path allows a client to retrieve a kubeconfig
file for a specified cluster reference of a logical cloud.
- includes CA cert, address, user private key and signed cert.
This commit includes the "lazy-loading" implementation of certificate
retrieval per cluster from Rsync (which happens when clients call).
The certificate is read from the cluster status in appcontext.
Thus, Monitor and Rsync need to be configured and running.
Issue-ID: MULTICLOUD-1143
Change-Id: Ie94cd128e14c8a944861eced2bdc886d95fab6ed
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
|
|
Make use of the new /subresource level (under a resource) to
request rsync to process approvals (a K8s subresource) for CSRs.
Generic usage of instructions (order, dependency) for consistency.
This also introduces a 'subresources' package in 'appcontext' with
the first member being the approval subresource. Since subresources
aren't necessarily created with yaml, this package will allow
developers to specify what should be the "interface" in appcontext
for each type of subresource when the different services need to
exchange this data (e.g. DCM and rsync).
Issue-ID: MULTICLOUD-1143
Change-Id: I8c85b4ee8c5654036a1c7ebb39d111ba4518b468
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
|
|
|
|
Before this patch, any resource quotas, say, with a dot (like
limits.cpu), would not be stored properly in the database and thus
unusable and ignored by DCM logical cloud apply operation.
Issue-ID: MULTICLOUD-1143
Change-Id: I178e66756bc9bb2798427233d15196d0e2559a99
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
"Rest of" meaning the Apply operation now effectively
instantiates logical cloud resources on K8s cluster.
More specifically, this commit now enables:
- DCM calling rsync over gRPC on apply
- The following K8s resources are now being created per cluster:
- namespace, csr, resourcequotas, roles, rolebindings
- Tested-working for multiple clusters simultaneously
- Generating and storing logical cloud user private key in MongoDB
This commit doesn't include any logic to request
or retrieve CSR approvals and signed user certificates.
Issue-ID: MULTICLOUD-1143
Change-Id: I92c4270678d05a5cb531791cef199a7dd79403d3
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
When applying, check whether logical cloud is already applied.
When deleting logical cloud, check if it exists.
Issue-ID: MULTICLOUD-1143
Change-Id: I6aa11d8921db76e3d9e9632f915b55fe7ab8efab
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
Also makes minor changes to non-terminate code as a side-effect of
supporting the new Terminate operation (such as including tagContext in
the LogicalCloudClient implementation of LogicalCloudManager interface).
These changes are/will also be leveraged by other operations.
Issue-ID: MULTICLOUD-1143
Change-Id: Idbd2ec9f6cf0e5584a0f51cf4c16144db56d9fa0
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
These changes allow the Monitor to also track CSR
(CertificateSigningResource) resources which will make
it possible to know when a certificate has been issued by
the K8s cluster signer. In turn, DCM will be able to read,
store and use that certificate to generate kubeconfigs.
Out-of-tree actions required:
- publish monitor's docker image built from this source
onto emcov2/monitor:latest
Issue-ID: MULTICLOUD-1143
Change-Id: I7facd27bbfe08891151bb3b6a9a19948435e24e4
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
|
|
And log error appropriately.
Essentially, add basic error handling and return from waitForDone.
It is possible to trigger this error easily by having an rsync client
wipe out etcd right after issuing an InvokeUninstallApp, as demonstrated
by DCM (before it was adapted to use the new asynchronous rsync).
Issue-ID: MULTICLOUD-1143
Change-Id: Id5c0d8da928738cb3cdc862eaefdb91db2eadf5f
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
This improves error handling between DCM and the database resources and
adds/corrects a lot more HTTP return codes in the API, respectively.
Issue-ID: MULTICLOUD-1143
Change-Id: I3abc8025660e042f4c946f8bbfd280e1eb4c9583
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
|
|
DCM listening on port 9077 as default.
The IP addresses for MongoDB and etcd are what we get by default
on docker installations of those using emco deployments.
Issue-ID: MULTICLOUD-1143
Change-Id: I9b30b278e3cf4fdb428665986e05c6916000b4e7
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
Instructions (order/dependency) can now belong to an
app, a resource or a subresource.
Issue-ID: MULTICLOUD-1143
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
Change-Id: Iee55c2bbb569de242e66f5a5456ac52025a00f0e
|
|
In this patch,we clean up the tmp files
and directories generated during the
process of instantiation.
They include the tmp dirs with
prefix : /tmp/helm-tmpl and /tmp/k8s-ext
Issue-ID: MULTICLOUD-1206
Signed-off-by: Rajamohan Raj <rajamohan.raj@intel.com>
Change-Id: I02d11bb2f8d920e35aae7343f041a53b1cd3f057
|
|
Signed-off-by: vikaskumar <vkumar@aarnanetworks.com>
Change-Id: I78b2f9f4ec3aa0b7a660a21e2ee2d61a3cddadc6
Issue-ID: MULTICLOUD-1225
|
|
goautoneg used to be available from bitbucket.org/ww/goautoneg
but not anymore. operator-sdk has been importing it from
github.com/munnerz/goautoneg for some time now but since we were
using operator-sdk v0.9.0, it was still pointing at the old (now
deleted) repository.
This patch bumps operator-sdk to v0.19.0 (which doesn't use
goautoneg from bitbucket), controller-runtime to v0.5.11, helm to
v2.16.12 and all k8s.io dependencies to v0.16.9 (consistent with
the k8s v1.16.9 we use).
Code changes have been made to make all modules compatible with
the new dependencies.
All src go.sums have been cleaned (deleted and re-built).
Issue-ID: MULTICLOUD-1143
Change-Id: I89fc39e9595b4a6a38dd90028b161bcecd3ef349
Signed-off-by: Igor D.C <igor.duarte.cardoso@intel.com>
|
|
Issue-ID: MULTICLOUD-1147
Signed-off-by: Aditya Sharoff <aditya.sharoff@intel.com>
Change-Id: I659dfcb00b3f15bcf787ac3947f8755737b1fc39
|
|
Issue-ID: MULTICLOUD-1179
Signed-off-by: Konrad Bańka <k.banka@samsung.com>
Change-Id: I2f8facefc5299408a47dd9f8bdba9410aa8171de
|
|
|
|
|
|
Issue-ID: MULTICLOUD-1198
Signed-off-by: Rajamohan Raj <rajamohan.raj@intel.com>
Change-Id: Idc2fbfd6e875bd7b662c7d60925aeb776b3f7bea
|
|
|
|
|
|
- Adds retry watcher and related functionality.
- Adds code to update, get the status from appcontext.
- Adds logic to handle state transition during terminate.
Issue-ID: MULTICLOUD-1005
Signed-off-by: Manjunath Ranganathaiah <manjunath.ranganathaiah@intel.com>
Change-Id: I2ed76efd9d8b6f40fec547bbe8b7d8a86f69ce07
|
|
In this patch, a new config item for log-level
is added, default log-level is set as "warn",
for detailed logs, set log-level as "info"
Issue-ID: MULTICLOUD-1200
Signed-off-by: Rajamohan Raj <rajamohan.raj@intel.com>
Change-Id: I3205ce110a492ecc6a7c680e3d35e173a5624bb0
|
|
Allow release-name property to be provided during instantiation that, if
provided, overrides release-name specified in profile.
Additionally updated Makefile to allow easy compilation with different
go version easily.
Issue-ID: MULTICLOUD-1175
Signed-off-by: Konrad Bańka <k.banka@samsung.com>
Change-Id: Id8db484369045cfb0bc99543a80317644fc838f9
|
|
Issue-ID: MULTICLOUD-1148
Signed-off-by: Aditya Sharoff <aditya.sharoff@intel.com>
Change-Id: I482cb8ff06ebf945020c5fc185479d9911aa946a
|
|
|
|
CLI updated with get functionality. Also adding
vFw test case
Issue-ID: MULTICLOUD-1065
Signed-off-by: Ritu Sood <ritu.sood@intel.com>
Change-Id: I7bd22aca9fac9cb7b1f4c93d0ffad5b07b62cced
|
|
Issue-ID: MULTICLOUD-1065
Signed-off-by: Ritu Sood <ritu.sood@intel.com>
Change-Id: I1ae50bafe7b4e40811d5d2061460e1c0240e3dde
|
|
|
|
|
|
This patch enhances the status query API.
- The ResourceBundleState CRD is modified to just use the
k8s Pod structure instead of a customized struct.
- Status queries can either present results showing
the rsync status of the composite app and resources
or from information received from the cluster via
the ResourceBundleState CR
- Query parameters are provided to the API call to
customize the query and response
- Support for querying status of cluster network
intents is added
Issue-ID: MULTICLOUD-1042
Signed-off-by: Eric Multanen <eric.w.multanen@intel.com>
Change-Id: Icca4cdd901e2f2b446414fade256fc24d87594cd
|
|
Changes made to remove dependency of rsyn
registraion from orchestrator need to be added
to ncm as well.
Issue-ID: MULTICLOUD-1196
Signed-off-by: Eric Multanen <eric.w.multanen@intel.com>
Change-Id: Ie572a127fd87cf50a1bb03163d98d3ebe901437e
|
|
Issue-ID: MULTICLOUD-1196
Signed-off-by: Rajamohan Raj <rajamohan.raj@intel.com>
Change-Id: Ib5b59d025151f53d64d089a649a5b25f572d2c60
|