Press "Enter" to skip to content

百度飞桨(PaddlePaddle)分布式训练在 Volcano 系统上的实践

 

 

来源 | 容器魔方公众号

 

飞桨 (PaddlePaddle) 是百度于 2016 年 9 月开源的深度学习框架,旨在提供一款安全高效、灵活易用、可扩展的深度学习平台。

 

2018 年 10 月,飞桨团队发布 Paddle Fluid 1.0 版本,对神经网络描述、大规模分布式训练、高性能推理引擎等核心能力进行了全面升级。以工业界应用必需的分布式训练能力为例,在最新的 Paddle Fluid 1.5.2 版本中,飞桨支持数据并行、模型并行、流水线并行等多种并行模式,参数服务器架构和点对点同步训练架构全面支持在 CPU、GPU 等硬件资源设备上的大规模训练。本文将介绍飞桨分布式训练在 Kubernetes 社区中的 Volcano 系统上进行实践的案例。

 

Kubernetes 是当今最火的容器化应用自动部署、伸缩和资源管理的开源系统。随着 Kubernetes 的崛起,越来越多的公司愿意将自己的业务应用部署在 Kubernetes 上。除了典型的 Web 服务、数据库等服务会基于 Kubernetes 进行部署以外,深度学习框架的分布式训练也不例外。

 

然而,在 Kubernetes 系统中提交深度学习训练任务的功能并不像传统的高性能计算 MPI 平台那样直观。在 2017 年,Kubernetes 社区就有文章 Run Deep Learning with PaddlePaddle on Kubernetes 分析了运行 PaddlePaddle 对底层资源的诉求,基于 PaddlePaddle 对计算容错性、弹性伸缩、资源隔离的要求,提出在 Kubernetes 平台上运行 PaddlePaddle 是最佳实践。

 

自 Paddle Fluid 1.0 发布以来,飞桨在平台部署和任务调度管理上已经取得了长足的进步。借助 Kubernetes 平台,飞桨可以实现 CPU/GPU 等硬件资源的合理调度分配、训练任务的弹性扩缩容,并能显着提升计算资源的利用率。但是,在并行创建和调度任务、训练任务的生命周期管理、计算资源亲和性调度、调度策略优化等方面还有提升空间。为了提升飞桨框架的计算效率,飞桨团队和 Volcano 团队联合发布 PaddlePaddle on Volcano 方案。

 

Volcano 是一款构建于 Kubernetes 之上的增强型高性能计算任务批量处理系统。

 

作为一个面向高性能计算场景的平台,它弥补了 kubernetes 在机器学习、深度学习、HPC、大数据计算等场景下的基本能力缺失,其中包括 gang-schedule 的调度能力、计算任务队列管理、GPU 亲和性调度。另外,Volcano 在原有 Kubernetes 能力基础上对计算任务的批量创建及生命周期管理、Fair-share 调度等方面做了增强。

 

Volcano 平台可以满足飞桨对资源创建,资源调度的基本要求。Volcano 的批量创建批量调度计算任务为飞桨作业提供计算任务的自动化生命周期管理,gang-scheduler 调度策略可以满足 PServer 和 Trainer “all or nothing”的业务调度约束,Queue 和 priority 逻辑可以管理集群下计算作业的执行顺序,Fair-share 和 GPU 亲和度调度使计算任务调度更贴合 PServer 和 Trainer 对节点资源和网络拓扑结构的要求而提升任务计算效率。

 

Volcano 借助 Kubernetes 创建 CRD 能力,在 Kubernetes 中引入“apiVersion”为“batch.volcano.sh/v1alpha1”,“kind”为“Job”的资源对象,用于描述计算任务。通过配置和创建 Volcano job 可以使用 Volcano 平台创建、管理和调度计算任务。使用 volcano 平台,需要先在 Kubernetes 集群下安装 Volcano,安装 Volcano 的方法可参考 Volcano 官网。

 

选择一个飞桨框架任务分别使用 Kubernetes 原生资源和 Volcano job 执行计算任务并对比分析,以下对比将着重体现两者在使用方法、任务管理、调度策略方面进行比较。选择飞桨官网分布式训练 CTR(Click-Through-Rate) demo 进行对比测试。CTR demo 将运行两个 PServer 任务和两个 Trainer 任务。

 

首先使用飞桨官网推荐模式执行分布式计算任务,先创建一个副本数为 2 的 Kubernetes ReplicaSet 对象,用于运行 PServer 业务,然后创建一个并行度为 2 的 Kubernetes Job 对象,用于运行 Trainer 任务。

 

创建 PServer 任务

 

root@volcano-paddlepaddle:~# kubectl apply -f pserver.yaml

replicaset.extensions/fluid-ctr-pserver create

 

查看 pserver ReplicaSet 组件

 

root@volcano-paddlepaddle:~# kubectl get rs

NAME DESIRED CURRENT READY AGE

fluid-ctr-pserver 2 2 2 5

 

查看 pserver pods

 

root@volcano-paddlepaddle:~# kubectl get pods | grep fluid

fluid-ctr-pserver-b9w99 1/1 Running 0 9m18s

fluid-ctr-pserver-pb9vd 1/1 Running 0 9m18

 

查看 pserver 日志,PServer 已经开始工作,并对外提供服务

 

root@volcano-paddlepaddle:~# kubectl logs fluid-ctr-pserver-b9w99

 

 

+ case “$1″in

+ start_fluid_process

+ pserver_label=paddle-job-pserver=fluid-ctr

+ trainer_label=paddle-job=fluid-ct

+ hostname=c-rlnrdybm-muamumvq-1

+ task_index=

+ ‘[‘ PSERVER == TRAINER ‘]

+ ‘[‘ PSERVER == PSERVER ‘]’

+ stdbuf -oL python /root/k8s_tools.py wait_pods_running paddle-job-pserver=fluid-ctr 2

label selector: paddle-job-pserver=fluid-ctr, desired: 2

current cnt: 0 sleep for 5 seconds…

+ ‘[‘ PSERVER == TRAINER ‘]’

+ ‘[‘ PSERVER == WORKER ‘]

++ python /root/k8s_tools.py fetch_endpoints paddle-job-pserver=fluid-ctr 30236

+ export PADDLE_PSERVERS=192.168.48.24:30236,192.168.48.25:30237

+ PADDLE_PSERVERS=192.168.48.24:30236,192.168.48.25:30237

++ python /root/k8s_tools.py fetch_ips paddle-job=fluid-ctr

+ export PADDLE_TRAINER_IPS=

+ PADDLE_TRAINER_IPS=

+ ‘[‘ PSERVER == TRAINER ‘]’

+ ‘[‘ PSERVER == WORKER ‘]’

++ python /root/k8s_tools.py fetch_id paddle-job-pserver=fluid-ctr

+ task_index=0

+ export PADDLE_TRAINER_ID=0

+ PADDLE_TRAINER_ID=0

+ export PADDLE_PSERVER_ID=0

+ PADDLE_PSERVER_ID=0

+ stdbuf -oL sh -c ‘cd /workspace/ctr && python train.py –is_local 0 –cloud_train 1’

2019-09-03 06:43:10,661 – INFO – run dist training

2019-09-03 06:43:10,715 – INFO – run pserver

get_pserver_program() is deprecated, call get_pserver_programs() to get pserver main and startup in a single call.

I0903 06:43:10.826609 41 grpc_server.cc:435] Server listening on 192.168.48.24:30236 selected port:

 

创建 trainer 任务

 

root@volcano-paddlepaddle:~# kubectl apply -f trainer.yaml

job.batch/fluid-ctr-trainer create

 

查看 trainer pods

 

root@volcano-paddlepaddle:~# kubectl get pod | grep fluid

fluid-ctr-pserver-b9w99 1/1 Running 0 87m

fluid-ctr-pserver-pb9vd 1/1 Running 0 87m

fluid-ctr-trainer-lg9n5 1/1 Running 0 12s

fluid-ctr-trainer-tvr99 1/1 Running 0 12

 

查看 trainer 任务日志,看到任务已经开始执行

 

root@volcano-paddlepaddle:~# kubectl logs fluid-ctr-trainer-lg9n5

 

+ case “$1” in

+ start_fluid_process

+ pserver_labe=paddle-job-pserver=fluid-ctr

+ trainer_label=paddle-job=fluid-ctr

+ hostname=c-rlnrdybm-muamumvq-2

+ task_index=

+ ‘[‘ TRAINER == TRAINER ‘]’

+ stdbuf -oL python /root/k8s_tools.py wait_pods_running paddle-job-pserver=fluid-ctr 2

label selector: paddle-job-pserver=fluid-ctr, desired: 2

+ ‘[‘ TRAINER == TRAINER ‘]’

+ stdbuf -oL python /root/k8s_tools.py wait_pods_running paddle-job=fluid-ctr 2

label selector: paddle-job=fluid-ctr, desired: 2

++ python /root/k8s_tools.py fetch_endpoints paddle-job-pserver=fluid-ctr 30236

+ export PADDLE_PSERVERS=192.168.48.24:30236,192.168.48.25:30237

+ PADDLE_PSERVERS=192.168.48.24:30236,192.168.48.25:30237

++ python /root/k8s_tools.py fetch_ips paddle-job=fluid-ctr

+ export PADDLE_TRAINER_IPS=192.168.48.24,192.168.48.25

+ PADDLE_TRAINER_IPS=192.168.48.24,192.168.48.25

+ ‘[‘ TRAINER == TRAINER ‘]’

+ check_failed_cnt 1

+ max_failed=1

++ python /root/k8s_tools.py count_pods_by_phase paddle-job=fluid-ctr Failed

+ failed_count=0

+ ‘[‘ 0-gt 1 ‘]’

++ python /root/k8s_tools.py fetch_id paddle-job=fluid-ctr

+ task_index=0

+ export PADDLE_TRAINER_ID=0

+ PADDLE_TRAINER_ID=0

+ export PADDLE_PSERVER_ID=0

+ PADDLE_PSERVER_ID=0

+ stdbuf -oL sh -c ‘cd /workspace/ctr && python train.py –is_local 0 –cloud_train 1’

2019-09-03 08:10:20,888 – INFO – run dist training

2019-09-03 08:10:20,951 – INFO – download the training materials

% Total % Received % Xferd Average Speed Time Time Time Current

Dload Upload Total Spent Left Speed

100 433M 100 433M 0 0 70.9M 0 0:00:06 0:00:06 –:–:– 97.0M

2019-09-03 08:11:04,522 – INFO – run trainer

2019-09-03 08:11:04,591 – WARNING –

I0903 08:11:04.594007 25 parallel_executor.cc:329] The number of CPUPlace, which is used in ParallelExecutor, is 2. And the Program will be copied 2 copies

I0903 08:11:04.875757 25 rpc_client.h:101] init rpc client with trainer_id 0

2019-09-03 08:11:38,625 – INFO – TRAIN –> pass: 0 batch: 0 loss: 0.697331115723 auc: 0.500826068453, batch_auc: 0.500826068453

2019-09-03 08:11:38,967 – INFO – TRAIN –> pass: 0 batch: 1 loss: 0.652093688965 auc: 0.48451329672, batch_auc: 0.48451329672

2019-09-03 08:11:39,242 – INFO – TRAIN –> pass: 0 batch: 2 loss: 0.629092956543 auc: 0.485173881519, batch_auc: 0.485173881519

2019-09-03 08:11:39,577 – INFO – TRAIN –> pass: 0 batch: 3 loss: 0.603850708008 auc: 0.482131778494, batch_auc: 0.482131778494

2019-09-03 08:11:39,874 – INFO – TRAIN –> pass: 0 batch: 4 loss: 0.591485412598 auc: 0.479737304993, batch_auc: 0.479737304993

2019-09-03 08:11:40,133 – INFO – TRAIN –> pass: 0 batch: 5 loss: 0.58376159668 auc: 0.478554220739, batch_auc: 0.478554220739

2019-09-03 08:11:40,385 – INFO – TRAIN –> pass: 0 batch: 6 loss: 0.561969116211 auc: 0.481465857424, batch_auc: 0.481465857424

2019-09-03 08:11:40,637 – INFO – TRAIN –> pass: 0 batch: 7 loss: 0.557065185547 auc: 0.486014931119, batch_auc: 0.486014931119

2019-09-03 08:11:40,890 – INFO – TRAIN –> pass: 0 batch: 8 loss: 0.562498413086 auc: 0.489651573333, batch_auc: 0.489651573333

2019-09-03 08:11:41,158 – INFO – TRAIN –> pass: 0 batch: 9 loss: 0.566428283691 auc: 0.489853260221, batch_auc: 0.49137884426

2019-09-03 08:11:41,452 – INFO – TRAIN –> pass: 0 batch: 10 loss: 0.564840087891 auc: 0.492880386228, batch_auc: 0.494013763938

2019-09-03 08:11:41,742 – INFO – TRAIN –> pass: 0 batch: 11 loss: 0.564809204102 auc: 0.493201528907, batch_auc: 0.498872381582

2019-09-03 08:11:42,056 – INFO – TRAIN –> pass: 0 batch: 12 loss: 0.584479736328 auc: 0.494151972036, batch_auc: 0.503926628391

2019-09-03 08:11:42,329 – INFO – TRAIN –> pass: 0 batch: 13 loss: 0.615677246094 auc: 0.49252557362, batch_auc: 0.5028352489

 

等待 trainer 任务执行完成,查看 pserver 和 trainer pods 状态,trainer 已经执行完成,pserver 仍然于运行中

 

root@volcano-paddlepaddle:~# kubectl get pods | grep fluid

fluid-ctr-pserver-b9w99 1/1 Running 0 177m

fluid-ctr-pserver-pb9vd 1/1 Running 0 177m

fluid-ctr-trainer-lg9n5 0/1 Completed 0 90m

fluid-ctr-trainer-tvr99 0/1 Completed 0 90

 

将上述计算任务迁移到 volcano 平台上进行测试。

 

Volcano 支持 Multi-pod jobs,拓展“tasks”字段,tasks 下可以定义多个 pod 描述,其中“replicas” 字段描述 task 将要生成的 pod 数量,“name”描述 task 名称,pod 名称将根据 task 名称生成。Template 字段与 kubernetes “podTemplate”一致。ctr 的 demo 中含有两个 task:“pserver”和“trainer”,每个 task 的 replicas 都是 2,将会创建两个 PServer 任务,两个 Trainer 任务。

 

使用 Volcano 调度器,在 job 的配置中需要指定“schedulerName”为“volcano”,如果 schedulerName 没有指定为“volcano”,job 下的任务调度将会使用 kubernetes 的默认调度器“default”调度器。

 

Volcano 通过指定“minAvailable”字段保证计算任务的 gang-scheduler 调度策略。“minAvailable”数值指明在对当前计算任务下的 pods 进行调度时,需保证多少计算任务都能够调度才会执行调度任务,“minAvailable”的数值需要小于或等于计算任务下的所有任务数量的总和。对于 PaddlePaddle 框架计算任务,只有当所有的 PServer 和 Trainer 任务都处于运行中,才能开始计算任务。因此对于飞桨计算任务,“minAvailable”的数值需要与计算任务下的所有计算任务总和相等。

 

对于使用飞桨分布式训练的应用,在计算过程中,如果 PServer 任务或者 Trainer 任务被驱逐或失败,PServer 和 Trainer 形成的计算集群将会失效,所有的 PServer 任务和 Trainer 任务都需要重启,以形成新的集群开始新的计算。Volcano 可以通过设置“policies”实现上述目的。设置“PodEvicted”事件对应“RestartJob”动作,设置“PodFailed”事件对应“RestartJob”动作,在设置了这两个“policies”之后,当计算任务被驱逐或者失败,所有的计算任务将会重启。

 

下面是使用 Volcano 平台执行 CTR 任务的配置 ctr-volcano.yaml,配置文件可从 Volcano 代码库获取

 

Volcano 代码仓库地址:

 

https://github.com/volcano-sh/volcano/blob/master/example/integrations/paddlepaddle/ctr-paddlepaddle-on-volcano.yaml

 

apiVersion: batch.volcano.sh/v1alpha1

kind: Job

metadata:

name: ctr-volcano

spec:

minAvailable: 4

schedulerName: volcano

policies:

– event: PodEvicted

action: RestartJob

– event: PodFailed

action: RestartJob

tasks:

– replicas: 2

name: pserver

template:

metadata:

labels:

paddle-job-pserver: fluid-ctr

spec:

imagePullSecrets:

– name: default-secret

volumes:

– hostPath:

path: /home/work/

type: “”

name: seqdata

containers:

– image: volcanosh/edlctr:v1

command:

– paddle_k8s

– start_fluid

imagePullPolicy: IfNotPresent

name: pserver

volumeMounts:

– mountPath: /mnt/seqdata

name: seqdata

resources:

limits:

cpu: 10

memory: 30Gi

ephemeral-storage: 10Gi

requests:

cpu: 1

memory: 100M

ephemeral-storage: 1Gi

env:

– name: GLOG_v

value: “0”

– name: GLOG_logtostderr

value: “1”

– name: TOPOLOGY

value: “”

– name: TRAINER_PACKAGE

value: /workspace

– name: NAMESPACE

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: metadata.namespace

– name: POD_IP

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: status.podIP

– name: POD_NAME

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: metadata.name

– name: PADDLE_CURRENT_IP

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: status.podIP

– name: PADDLE_JOB_NAME

value: fluid-ctr

– name: PADDLE_IS_LOCAL

value: “0”

– name: PADDLE_TRAINERS_NUM

value: “2”

– name: PADDLE_PSERVERS_NUM

value: “2”

– name: FLAGS_rpc_deadline

” value: ”

– name: ENTRY

value: cd /workspace/ctr && python train.py –is_local 0 –cloud_train 1

– name: PADDLE_PORT

value: “30236”

– name: LD_LIBRARY_PATH

value: /usr/local/lib:/usr/local/nvidia/lib64:/usr/local/rdma/lib64:/usr/lib64/mlnx_ofed/valgrind

– name: PADDLE_TRAINING_ROLE

value: PSERVER

– name: TRAINING_ROLE

value: PSERVER

restartPolicy: OnFailure

– replicas: 2

policies:

– event: TaskCompleted

action: CompleteJob

name: trainer

template:

metadata:

labels:

paddle-job: fluid-ctr

spec:

imagePullSecrets:

– name: default-secret

volumes:

– hostPath:

path: /home/work/

type: “”

name: seqdata

containers:

– image: volcanosh/edlctr:v1

command:

– paddle_k8s

– start_fluid

imagePullPolicy: IfNotPresent

name: trainer

volumeMounts:

– mountPath: /mnt/seqdata

name: seqdata

resources:

limits:

cpu: 10

memory: 30Gi

ephemeral-storage: 10Gi

requests:

cpu: 1

memory: 100M

ephemeral-storage: 10Gi

env:

– name: GLOG_v

value: “0”

– name: GLOG_logtostderr

value: “1”

– name: TOPOLOGY

– name: TRAINER_PACKAGE

value: /workspace

– name: CPU_NUM

value: “2”

– name: NAMESPACE

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: metadata.namespace

– name: POD_IP

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: status.podIP

– name: POD_NAME

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: metadata.name

– name: PADDLE_CURRENT_IP

valueFrom:

fieldRef:

apiVersion: v1

fieldPath: status.podIP

– name: PADDLE_JOB_NAME

value: fluid-ctr

– name: PADDLE_IS_LOCAL

value: “0”

– name: FLAGS_rpc_deadline

” value: ”

– name: PADDLE_PORT

value: “30236”

– name: PADDLE_PSERVERS_NUM

value: “2”

– name: PADDLE_TRAINERS_NUM

value: “2”

– name: PADDLE_TRAINING_ROLE

value: TRAINER

– name: TRAINING_ROLE

value: TRAINER

– name: LD_LIBRARY_PATH

value: /usr/local/lib:/usr/local/nvidia/lib64:/usr/local/rdma/lib64:/usr/lib64/mlnx_ofed/valgrind

– name: ENTRY

value: cd /workspace/ctr && python train.py –is_local 0 –cloud_train 1

restartPolicy: OnFailure

 

在集群终端中执行以下指令在 default namespace 下创建 volcano job

 

root@volcano-paddlepaddle:~# kubectl apply -f ctr-volcano.yaml

job.batch.volcano.sh/ctr-volcano create

 

检查 pods 的状态,无论是 pserver 任务还是 trainer 任务都被下发到集群中,并开始运行。如果当前集群下的空闲资源,不能满足 pserver 任务和 trainer 任务的资源述求,任何任务都不会被创建。

 

root@volcano-paddlepaddle:~# kubectl get pods | grep ctr-volcano

ctr-volcano-pserver-0 1/1 Running 0 16s

ctr-volcano-pserver-1 1/1 Running 0 16s

ctr-volcano-trainer-0 1/1 Running 0 16s

ctr-volcano-trainer-1 1/1 Running 0 16

 

选择一个 PServer 任务查看日志,看到 PServer 在监听端口,并对外提供服

 

root@volcano-paddlepaddle:~# kubectl logs ctr-volcano-pserver-0

 

 

+ case “$1” in

+ start_fluid_process

+ pserver_label=paddle-job-pserver=fluid-ctr

+ trainer_label=paddle-job=fluid-ctr

+ hostname=ctr-volcano-pserver-0

+ task_index=

+ ‘[‘ PSERVER == TRAINER ‘]’

+ ‘[‘ PSERVER == PSERVER ‘]’

+ stdbuf -oL python /root/k8s_tools.py wait_pods_running paddle-job-pserver=fluid-ctr 2

label selector: paddle-job-pserver=fluid-ctr, desired: 2

current cnt: 0 sleep for 5 seconds…

+ ‘[‘ PSERVER == TRAINER ‘]’

+ ‘[‘ PSERVER == WORKER ‘]’

++ python /root/k8s_tools.py fetch_endpoints paddle-job-pserver=fluid-ctr 30236

+ export PADDLE_PSERVERS=172.20.0.148:30236,172.20.1.134:30237

+ PADDLE_PSERVERS=172.20.0.148:30236,172.20.1.134:30237

++ python /root/k8s_tools.py fetch_ips paddle-job=fluid-ctr

+ export PADDLE_TRAINER_IPS=172.20.0.147,172.20.1.133

+ PADDLE_TRAINER_IPS=172.20.0.147,172.20.1.133

+ ‘[‘ PSERVER == TRAINER ‘]’

+ ‘[‘ PSERVER == WORKER ‘]’

++ python /root/k8s_tools.py fetch_id paddle-job-pserver=fluid-ctr

+ task_index=0

+ export PADDLE_TRAINER_ID=0

+ PADDLE_TRAINER_ID=0

+ export PADDLE_PSERVER_ID=0

+ PADDLE_PSERVER_ID=0

+ stdbuf -oL sh -c ‘cd /workspace/ctr && python train.py –is_local 0 –cloud_train 1’

2019-09-03 09:57:55,619 – INFO – run dist training

2019-09-03 09:57:55,708 – INFO – run pserver

get_pserver_program() is deprecated, call get_pserver_programs() to get pserver main and startup in a single call.

I0903 09:57:55.860916 41 grpc_server.cc:435] Server listening on 172.20.0.148:30236 selected port:

 

选择一个 Trainer 任务查看日志,看到计算任务已经开始执行

 

root@volcano-paddlepaddle:~# kubectl logs ctr-volcano-trainer-0

 

 

+ case “$1” in

+ start_fluid_process

+ pserver_label=paddle-job-pserver=fluid-ctr

+ trainer_label=paddle-job=fluid-ctr

+ hostname=ctr-volcano-trainer-0

+ task_index=

+ ‘[‘ TRAINER == TRAINER ‘]’

+ stdbuf -oL python /root/k8s_tools.py wait_pods_running paddle-job-pserver=fluid-ctr 2

label selector: paddle-job-pserver=fluid-ctr, desired: 2

current cnt: 0 sleep for 5 seconds…

+ ‘[‘ TRAINER == TRAINER ‘]’

+ stdbuf -oL python /root/k8s_tools.py wait_pods_running paddle-job=fluid-ctr 2

label selector: paddle-job=fluid-ctr, desired: 2

++ python /root/k8s_tools.py fetch_endpoints paddle-job-pserver=fluid-ctr 30236

+ export PADDLE_PSERVERS=172.20.0.148:30236,172.20.1.134:30237

+ PADDLE_PSERVERS=172.20.0.148:30236,172.20.1.134:30237

++ python /root/k8s_tools.py fetch_ips paddle-job=fluid-ctr

+ export PADDLE_TRAINER_IPS=172.20.0.147,172.20.1.133

+ PADDLE_TRAINER_IPS=172.20.0.147,172.20.1.133

+ ‘[‘ TRAINER == TRAINER ‘]’

+ check_failed_cnt 1

+ max_failed=1

++ python /root/k8s_tools.py count_pods_by_phase paddle-job=fluid-ctr Failed

+ failed_count=0

+ ‘[‘ 0 -gt 1 ‘]’

++ python /root/k8s_tools.py fetch_id paddle-job=fluid-ctr

+ task_index=0

+ export PADDLE_TRAINER_ID=0

+ PADDLE_TRAINER_ID=0

+ export PADDLE_PSERVER_ID=0

+ PADDLE_PSERVER_ID=0

+ stdbuf -oL sh -c ‘cd /workspace/ctr && python train.py –is_local 0 –cloud_train 1’

2019-09-03 09:57:56,712 – INFO – run dist training

2019-09-03 09:57:56,773 – INFO – download the training materials

% Total % Received % Xferd Average Speed Time Time Time Current

Dload Upload Total Spent Left Speed

100 433M 100 433M 0 0 96.2M 0 0:00:04 0:00:04 –:–:– 96.2M

2019-09-03 09:58:27,648 – INFO – run trainer

2019-09-03 09:58:27,732 – WARNING –

I0903 09:58:27.734141 25 parallel_executor.cc:329] The number of CPUPlace, which is used in ParallelExecutor, is 2. And the Program will be copied 2 copies

I0903 09:58:27.937546 25 rpc_client.h:101] init rpc client with trainer_id 0

2019-09-03 09:58:37,957 – INFO – TRAIN –> pass: 0 batch: 0 loss: 0.670620727539 auc: 0.510430537062, batch_auc: 0.510764985415

2019-09-03 09:58:38,264 – INFO – TRAIN –> pass: 0 batch: 1 loss: 0.641319274902 auc: 0.503955813399, batch_auc: 0.503955813399

2019-09-03 09:58:38,585 – INFO – TRAIN –> pass: 0 batch: 2 loss: 0.617138793945 auc: 0.50334993182, batch_auc: 0.50334993182

2019-09-03 09:58:38,873 – INFO – TRAIN –> pass: 0 batch: 3 loss: 0.598490356445 auc: 0.507263818365, batch_auc: 0.507263818365

2019-09-03 09:58:39,182 – INFO – TRAIN –> pass: 0 batch: 4 loss: 0.573976501465 auc: 0.510442316749, batch_auc: 0.51044231674

 

等待大概 70 分钟,查看计算任务日志,发现任务已经安全退出

 

root@volcano-paddlepaddle:~# kubectl get pod | grep ctr-volcano

ctr-volcano-trainer-0 0/1 Completed 0 77m

ctr-volcano-trainer-1 0/1 Completed 0 77

 

与此同时,在训练结束之后,我们可能需要训练出来的模型用于别处。在 yaml 文件当中,我们规定了该任务 volcanosh/edlctr:v1 镜像,该镜像的工作目录在 /workspace/ctr 下,在 train.py 当中有定义,会在每 1000 个 batch 或是每一轮 pass(跑完一遍训练集)的时候,调用 save_inference_model 接口来保存模型。保存的模型在 /workspace/ctr/models 文件夹下。那幺如何在任务结束后获取模型呢?我们建议以下几种方式。

 

 

在 yaml 文件当中 trainer 部分的 spec 当中定义 volume,通过 docker 的 volume 映射容器路径和宿主机路径的机制,将 /workspace/ctr/models 文件夹映射到宿主机的文件夹中。接下来通过 kubectl describe pod ctr-volcano-trainer-0,可以得知我们的模型所在的节点,接下来 ssh 登陆到对应的节点上,到宿主机被映射到路径下,就可以获取到训练出来到模型了。

 

如果需要更加灵活的,自动化的模型配送流程,可以在 K8S 集群上建立 File Server 和分布式文件系统,例如 GlusterFS。将 ctr-volcano-trainer-0 容器内部的 /workspace/ctr/models 文件夹映射到 GlusterFS 的 PVC(Persistent Volume Claim)上。通过 ftp 的 wget/curl 操作命令就可以实现模型的获取和配送。

 

 

综上,使用 Volcano 平台执行 PaddlePaddle 框架计算任务,可以实现计算任务的批量创建,任务的自动化管理,实现计算任务的自我管理。相较于普通的 Replicaset+Job 的模式,使用 Volcano 平台可以提升并行计算的管理效率。

 

作者介绍

 

董大祥
,@guru4elephant, PaddlePaddle Architect, Principal Architect, Baidu

 

王嘉炜
,@wangjiawei04, PaddlePaddle Engineer, Senior Engineer, Baidu

 

于佃海
,@raindrops2sea, PaddlePaddle Architect, Distinguished Architect, Baidu

 

张经辉
,@sivanzcw, Volcano Contributor, Cloud Native software engineer, Huawei

 

马达
,@k82cn, Kubernetes Maintainer, SIG-Scheduling Co-Leader, Volcano Lead, Huawei

 

参考文献:

 

PaddlePaddle 官网

 

https://www.paddlepaddle.org.cn

 

Paddle on Spark

 

https://github.com/hohdiy/paddle_on_spark/blob/master/doc/paddle_on_spark.md

 

Run Deep Learning with PaddlePaddle on Kubernetes

 

https://kubernetes.io/blog/2017/02/run-deep-learning-with-paddlepaddle-on-kubernetes/

 

Volcano 官网

 

https://volcano.sh

 

Volcao 社区

 

https://github.com/volcano-sh/volcano

 

百度 CTR Demo

 

https://www.paddlepaddle.org.cn/documentation/docs/zh/1.5/user_guides/howto/training/deploy_ctr_on_baidu_cloud_cn.html

 

CTR-volcano 配置文件

 

https://github.com/volcano-sh/volcano/blob/master/example/integrations/paddlepaddle/ctr-paddlepaddle-on-volcano.yam

Be First to Comment

发表回复

您的电子邮箱地址不会被公开。 必填项已用*标注