tfserving部署

1. 进入root 用户下
   1. yum -y install docker 安装容器
   2. systemctl start docker 启动容器
   
   1. 查看命令 sudo docker ps ,如图所示 ，说明启动成功
   
   1. 拉取镜像 sudo docker pull tensorflow/serving:2.11.0 如图
   
   1. 查看镜像 sudo docker images
   

 

docker的文件路径：

配置文件和模型文件

 

 

2. docker的启动脚本 run.sh

?

1 2 3 4 5 6 7 8 9

#!/bin/bash   sudo docker run -t --rm -p 8016:8500 -p 8017:8501 --name prerank_8016  --cpuset-cpus="64-71" --memory="16g" -v "/home/public_sudo/data/prerank:/models" -v "/ home/public_sudo/data/prerank/config:/config"  -e TF_CPP_MIN_VLOG_LEVEL=4 tensorflow/serving:2.11.0  --per_process_gpu_memory_fraction=0.8  --enable_batching =true --model_config_file_poll_wait_seconds=60 --model_config_file=/config/model.config > /home/public_sudo/data/prerank/logs/log8016.log &   sudo docker run -t --rm -p 8018:8500 -p 8019:8501 --cpuset-cpus="72-79" --memory="16g" -v "/home/public_sudo/data/prerank:/models" -v "/home/public_sudo/data /prerank/config:/config"  --name prerank_8018 -e TF_CPP_MIN_VLOG_LEVEL=4 tensorflow/serving:2.11.0  --per_process_gpu_memory_fraction=0.8  --enable_batching= true --model_config_file_poll_wait_seconds=60 --model_config_file=/config/model.config > /home/public_sudo/data/prerank/logs/log8018.log &<br><br><br>

sudo docker run -t --rm -p 8020:8500 -p 8021:8501 --name relevance_8020 --cpuset-cpus="80-87" --memory="16g" -v "/home/public_sudo/data/relevance:/models" -v "/home/public_sudo/data/relevance/config:/config" -e TF_CPP_MIN_VLOG_LEVEL=4 tensorflow/serving:2.11.0 --per_process_gpu_memory_fraction=0.8 --enable_batching=true --model_config_file_poll_wait_seconds=60 --model_config_file=/config/model.config > /home/public_sudo/data/relevance/logs/8020.log &

sudo docker run -t --rm -p 8022:8500 -p 8023:8501 --name relevance_8022 --cpuset-cpus="88-95" --memory="16g" -v "/home/public_sudo/data/relevance:/models" -v "/home/public_sudo/data/relevance/config:/config" -e TF_CPP_MIN_VLOG_LEVEL=4 tensorflow/serving:2.11.0 --per_process_gpu_memory_fraction=0.8 --enable_batching=true --model_config_file_poll_wait_seconds=60 --model_config_file=/config/model.config > /home/public_sudo/data/relevance/logs/8022.log &

3. docker的配置文件格式

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53

model_config_list:{   config:{   name:'relevance_v5'   base_path:'/models/relevance_v5'   model_platform:'tensorflow'   },   config:{   name:'relevance_v4'   base_path:'/models/relevance_v4'   model_platform:'tensorflow'   },   config:{   name:'relevance_v3'   base_path:'/models/relevance_v3'   model_platform:'tensorflow'   },   config:{   name:'relevance_v1'   base_path:'/models/relevance_v1'   model_platform:'tensorflow'   },   config:{   name:'relevance_v2'   base_path:'/models/relevance_v2'   model_platform:'tensorflow'   }   }

4. docker的python测试

?

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120

import json import requests   input = [1 ,1 ,1 ,1 ,1 ,1 ,0 ,1 ,0.999 ,0 ,3 ,1 ,0.3333333333333333 ,2 ,0.6666666666666666 ,1 ,0.3333333333333333 ,0          ,0.0 ,0 ,0 ,0 ,1 ,0 ,1 ,0 ,0 ,1 ,0 ,1 ,0 ,0.8652462791972874 ,0.020301 ,0.626570119698537 ,0.0 ,1 ,1 ,1.0 ,0          ,0.0 ,1 ,1.0 ,0 ,0.0 ,1 ,1 ,0 ,0 ,0 ,1 ,0 ,0 ,0 ,0 ,0 ,1 ,0.8342278053174881 ,0.01272 ,1.0 ,0.0          ,0.08677808000000002 ,-0.5144478466666667 ,0.09813555641 ,0.4995743833333333 ,0.6875768999999999 ,0.29120327          ,-0.28226265 ,-0.4784384366666667 ,-0.58632582 ,0.04208295999999998 ,-0.32798395333333336 ,-0.02741264333333332          ,0.049279349666666666 ,-0.16968005666666663 ,-0.33240925466666665 ,0.4360717 ,-0.2929261233333333          ,-0.2981464466666667 ,-0.07636033 ,-0.274444911 ,-0.3359526518666667 ,0.50867007 ,0.39146848666666667          ,-0.7227929266666666 ,-0.191814214 ,0.043227369999999994 ,-0.8780203333333333 ,0.10589143999999999 ,0.15558687          ,-0.4661960733333333 ,-0.37050828333333335 ,0.04813903993333333 ,-0.142883988375 ,-0.734637185625          ,0.15876262870312502 ,0.7311148107500001 ,0.4145796666875001 ,-0.009326533625000001 ,-0.3222952091875          ,-0.5682045862499999 ,-0.8492591543749999 ,-0.057231256250000015 ,-0.2482288354999999 ,0.24096290889374997          ,-0.030687868800000004 ,-0.496588725 ,-0.29752982512500004 ,0.283379688875 ,-0.5157605628124999          ,-0.29660309737499996 ,0.23857122374999998 ,-0.2940595146875 ,-0.312397124 ,0.4648682387500001 ,0.3255289983125          ,-0.599876252 ,0.06666920775 ,0.2970209875 ,-0.48590265374999997 ,-0.20199137175 ,0.18776840000000006          ,-0.37830229231250007 ,-0.31018528725000005 ,0.02904392187500001]     def v1():     instance = []     instance.append({"relevance_input": input})     instance.append({"relevance_input": input})     data = json.dumps({"signature_name": "serving_default", "instances": instance})     headers = {"content-type": "application/json"}     json_response = requests.post('http://localhost:8021/v1/models/relevance_v1:predict',                                   data=data, headers=headers)     predictions = json.loads(json_response.text)["predictions"]       json_response1 = requests.post('http://localhost:8023/v1/models/relevance_v1:predict',                                   data=data, headers=headers)     predictions1 = json.loads(json_response1.text)["predictions"]     print("v1:", predictions, predictions1)     def v2():     instance = []     instance.append({"relevance_input": input[:60]})     instance.append({"relevance_input": input[:60]})     data = json.dumps({"signature_name": "serving_default", "instances": instance})     headers = {"content-type": "application/json"}     json_response = requests.post('http://localhost:8021/v1/models/relevance_v2:predict',                                   data=data, headers=headers)     predictions = json.loads(json_response.text)["predictions"]       json_response1 = requests.post('http://localhost:8023/v1/models/relevance_v2:predict',                                    data=data, headers=headers)     predictions1 = json.loads(json_response1.text)["predictions"]     print("v2:", predictions, predictions1)     def v3():     instance = []     instance.append({"relevance_input": input[:60]})     instance.append({"relevance_input": input[:60]})     data = json.dumps({"signature_name": "serving_default", "instances": instance})     headers = {"content-type": "application/json"}     json_response = requests.post('http://localhost:8021/v1/models/relevance_v3:predict',                                   data=data, headers=headers)     predictions = json.loads(json_response.text)["predictions"]       json_response1 = requests.post('http://localhost:8023/v1/models/relevance_v3:predict',                                    data=data, headers=headers)     predictions1 = json.loads(json_response1.text)["predictions"]     print("v3:", predictions, predictions1)     def v4():     instance = []     instance.append({"relevance_input": input[:60]})     instance.append({"relevance_input": input[:60]})     data = json.dumps({"signature_name": "serving_default", "instances": instance})     headers = {"content-type": "application/json"}     json_response = requests.post('http://localhost:8021/v1/models/relevance_v4:predict',                                   data=data, headers=headers)     predictions = json.loads(json_response.text)["predictions"]       json_response1 = requests.post('http://localhost:8023/v1/models/relevance_v4:predict',                                    data=data, headers=headers)     predictions1 = json.loads(json_response1.text)["predictions"]     print("v4:", predictions, predictions1)     def v5():     instance = []     instance.append({"relevance_input": input})     instance.append({"relevance_input": input})     data = json.dumps({"signature_name": "serving_default", "instances": instance})     headers = {"content-type": "application/json"}     json_response = requests.post('http://localhost:8021/v1/models/relevance_v5:predict',                                   data=data, headers=headers)     predictions = json.loads(json_response.text)["predictions"]       json_response1 = requests.post('http://localhost:8023/v1/models/relevance_v5:predict',                                    data=data, headers=headers)     predictions1 = json.loads(json_response1.text)["predictions"]     print("v5:", predictions, predictions1)   def v6():     instance = []     instance.append({"relevance_input": input})     instance.append({"relevance_input": input})     data = json.dumps({"signature_name": "serving_default", "instances": instance})     headers = {"content-type": "application/json"}     json_response = requests.post('http://localhost:8021/v1/models/relevance_v6:predict',                                   data=data, headers=headers)     predictions = json.loads(json_response.text)["predictions"]       json_response1 = requests.post('http://localhost:8023/v1/models/relevance_v6:predict',                                    data=data, headers=headers)     predictions1 = json.loads(json_response1.text)["predictions"]     print("v6:", predictions, predictions1)   if __name__ == '__main__':     v1()     v2()     v3()     v4()     v5()