Skip to content
GitLab
Menu
Projects
Groups
Snippets
Loading...
Help
Help
Support
Community forum
Keyboard shortcuts
?
Submit feedback
Contribute to GitLab
Sign in / Register
Toggle navigation
Menu
Open sidebar
OpenDAS
nni
Commits
14a79654
Unverified
Commit
14a79654
authored
May 23, 2019
by
SparkSnail
Committed by
GitHub
May 23, 2019
Browse files
Support v1beta2 operator in KubeflowTrainingService (#1058)
parent
1ace9baf
Changes
4
Hide whitespace changes
Inline
Side-by-side
Showing
4 changed files
with
102 additions
and
15 deletions
+102
-15
src/nni_manager/config/kubeflow/pytorchjob-crd-v1beta2.json
src/nni_manager/config/kubeflow/pytorchjob-crd-v1beta2.json
+17
-0
src/nni_manager/config/kubeflow/tfjob-crd-v1beta2.json
src/nni_manager/config/kubeflow/tfjob-crd-v1beta2.json
+17
-0
src/nni_manager/training_service/kubernetes/kubeflow/kubeflowApiClient.ts
...training_service/kubernetes/kubeflow/kubeflowApiClient.ts
+67
-14
src/nni_manager/training_service/kubernetes/kubeflow/kubeflowConfig.ts
...er/training_service/kubernetes/kubeflow/kubeflowConfig.ts
+1
-1
No files found.
src/nni_manager/config/kubeflow/pytorchjob-crd-v1beta2.json
0 → 100644
View file @
14a79654
{
"kind"
:
"CustomResourceDefinition"
,
"spec"
:
{
"scope"
:
"Namespaced"
,
"version"
:
"v1beta2"
,
"group"
:
"kubeflow.org"
,
"names"
:
{
"kind"
:
"PyTorchJob"
,
"plural"
:
"pytorchjobs"
,
"singular"
:
"pytorchjob"
}
},
"apiVersion"
:
"apiextensions.k8s.io/v1beta2"
,
"metadata"
:
{
"name"
:
"pytorchjobs.kubeflow.org"
}
}
src/nni_manager/config/kubeflow/tfjob-crd-v1beta2.json
0 → 100644
View file @
14a79654
{
"kind"
:
"CustomResourceDefinition"
,
"spec"
:
{
"scope"
:
"Namespaced"
,
"version"
:
"v1beta2"
,
"group"
:
"kubeflow.org"
,
"names"
:
{
"kind"
:
"TFJob"
,
"plural"
:
"tfjobs"
,
"singular"
:
"tfjob"
}
},
"apiVersion"
:
"apiextensions.k8s.io/v1beta2"
,
"metadata"
:
{
"name"
:
"tfjobs.kubeflow.org"
}
}
src/nni_manager/training_service/kubernetes/kubeflow/kubeflowApiClient.ts
View file @
14a79654
...
@@ -29,20 +29,35 @@ abstract class KubeflowOperatorClient extends KubernetesCRDClient{
...
@@ -29,20 +29,35 @@ abstract class KubeflowOperatorClient extends KubernetesCRDClient{
*/
*/
public
static
generateOperatorClient
(
kubeflowOperator
:
KubeflowOperator
,
public
static
generateOperatorClient
(
kubeflowOperator
:
KubeflowOperator
,
operatorApiVersion
:
string
):
KubernetesCRDClient
{
operatorApiVersion
:
string
):
KubernetesCRDClient
{
if
(
kubeflowOperator
===
'
tf-operator
'
)
{
switch
(
kubeflowOperator
)
{
if
(
operatorApiVersion
==
'
v1alpha2
'
)
{
case
'
tf-operator
'
:
{
return
new
TFOperatorClientV1Alpha2
();
switch
(
operatorApiVersion
)
{
}
else
if
(
operatorApiVersion
==
'
v1beta1
'
)
{
case
'
v1alpha2
'
:
{
return
new
TFOperatorClientV1Beta1
();
return
new
TFOperatorClientV1Alpha2
();
}
}
}
else
if
(
kubeflowOperator
===
'
pytorch-operator
'
)
{
case
'
v1beta1
'
:
{
if
(
operatorApiVersion
==
'
v1alpha2
'
)
{
return
new
TFOperatorClientV1Beta1
();
return
new
PytorchOperatorClientV1Alpha2
();
}
}
else
if
(
operatorApiVersion
==
'
v1beta1
'
)
{
case
'
v1beta2
'
:
{
return
new
PytorchOperatorClientV1Beta1
();
return
new
TFOperatorClientV1Beta2
();
}
}
break
;
}
}
case
'
pytorch-operator
'
:
{
switch
(
operatorApiVersion
)
{
case
'
v1alpha2
'
:
{
return
new
PyTorchOperatorClientV1Alpha2
();
}
case
'
v1beta1
'
:
{
return
new
PyTorchOperatorClientV1Beta1
();
}
case
'
v1beta2
'
:
{
return
new
PyTorchOperatorClientV1Beta2
();
}
}
}
}
}
throw
new
Error
(
`Invalid operator
${
kubeflowOperator
}
or apiVersion
${
operatorApiVersion
}
`
);
throw
new
Error
(
`Invalid operator
${
kubeflowOperator
}
or apiVersion
${
operatorApiVersion
}
`
);
}
}
}
}
...
@@ -85,7 +100,26 @@ class TFOperatorClientV1Beta1 extends KubernetesCRDClient {
...
@@ -85,7 +100,26 @@ class TFOperatorClientV1Beta1 extends KubernetesCRDClient {
}
}
}
}
class
PytorchOperatorClientV1Alpha2
extends
KubeflowOperatorClient
{
class
TFOperatorClientV1Beta2
extends
KubernetesCRDClient
{
/**
* constructor, to initialize tfjob CRD definition
*/
public
constructor
()
{
super
();
this
.
crdSchema
=
JSON
.
parse
(
fs
.
readFileSync
(
'
./config/kubeflow/tfjob-crd-v1beta2.json
'
,
'
utf8
'
));
this
.
client
.
addCustomResourceDefinition
(
this
.
crdSchema
);
}
protected
get
operator
():
any
{
return
this
.
client
.
apis
[
"
kubeflow.org
"
].
v1beta2
.
namespaces
(
'
default
'
).
tfjobs
;
}
public
get
containerName
():
string
{
return
'
tensorflow
'
;
}
}
class
PyTorchOperatorClientV1Alpha2
extends
KubeflowOperatorClient
{
/**
/**
* constructor, to initialize tfjob CRD definition
* constructor, to initialize tfjob CRD definition
*/
*/
...
@@ -104,7 +138,7 @@ class PytorchOperatorClientV1Alpha2 extends KubeflowOperatorClient {
...
@@ -104,7 +138,7 @@ class PytorchOperatorClientV1Alpha2 extends KubeflowOperatorClient {
}
}
}
}
class
Py
t
orchOperatorClientV1Beta1
extends
KubernetesCRDClient
{
class
Py
T
orchOperatorClientV1Beta1
extends
KubernetesCRDClient
{
/**
/**
* constructor, to initialize tfjob CRD definition
* constructor, to initialize tfjob CRD definition
*/
*/
...
@@ -123,5 +157,24 @@ class PytorchOperatorClientV1Beta1 extends KubernetesCRDClient {
...
@@ -123,5 +157,24 @@ class PytorchOperatorClientV1Beta1 extends KubernetesCRDClient {
}
}
}
}
class
PyTorchOperatorClientV1Beta2
extends
KubernetesCRDClient
{
/**
* constructor, to initialize tfjob CRD definition
*/
public
constructor
()
{
super
();
this
.
crdSchema
=
JSON
.
parse
(
fs
.
readFileSync
(
'
./config/kubeflow/pytorchjob-crd-v1beta2.json
'
,
'
utf8
'
));
this
.
client
.
addCustomResourceDefinition
(
this
.
crdSchema
);
}
protected
get
operator
():
any
{
return
this
.
client
.
apis
[
"
kubeflow.org
"
].
v1beta2
.
namespaces
(
'
default
'
).
pytorchjobs
;
}
public
get
containerName
():
string
{
return
'
pytorch
'
;
}
}
export
{
KubeflowOperatorClient
,
GeneralK8sClient
};
export
{
KubeflowOperatorClient
,
GeneralK8sClient
};
src/nni_manager/training_service/kubernetes/kubeflow/kubeflowConfig.ts
View file @
14a79654
...
@@ -28,7 +28,7 @@ import { MethodNotImplementedError } from '../../../common/errors';
...
@@ -28,7 +28,7 @@ import { MethodNotImplementedError } from '../../../common/errors';
export
type
KubeflowOperator
=
'
tf-operator
'
|
'
pytorch-operator
'
;
export
type
KubeflowOperator
=
'
tf-operator
'
|
'
pytorch-operator
'
;
export
type
DistTrainRole
=
'
worker
'
|
'
ps
'
|
'
master
'
;
export
type
DistTrainRole
=
'
worker
'
|
'
ps
'
|
'
master
'
;
export
type
KubeflowJobStatus
=
'
Created
'
|
'
Running
'
|
'
Failed
'
|
'
Succeeded
'
;
export
type
KubeflowJobStatus
=
'
Created
'
|
'
Running
'
|
'
Failed
'
|
'
Succeeded
'
;
export
type
OperatorApiVersion
=
'
v1alpha2
'
|
'
v1beta1
'
;
export
type
OperatorApiVersion
=
'
v1alpha2
'
|
'
v1beta1
'
|
'
v1beta2
'
;
export
class
KubeflowClusterConfig
extends
KubernetesClusterConfig
{
export
class
KubeflowClusterConfig
extends
KubernetesClusterConfig
{
public
readonly
operator
:
KubeflowOperator
;
public
readonly
operator
:
KubeflowOperator
;
...
...
Write
Preview
Markdown
is supported
0%
Try again
or
attach a new file
.
Attach a file
Cancel
You are about to add
0
people
to the discussion. Proceed with caution.
Finish editing this message first!
Cancel
Please
register
or
sign in
to comment