config.yaml Parameters

Parameter

Default Value

Description

cert.ca

None

Path of the CA certificate.

cert.crt

None

Path of the server certificate.

cert.key

None

Path of the private key of the server certificate.

cert.key_mm

None

Private key password of the server certificate.

cert.crl

None

Certificate revocation list (CRL).

check_cert_holder.enable

true

Whether to enable certificate holder verification. By default, certificate holder verification is enabled.

check_cert_holder.host_name

*.huawei.com

Verifies certificate validity. The DNS in the holder information Common Name (CN) and subjectAltName must match *.huawei.com. Do not to change the value.

upload_image_and_result

APULIS_ENDPOINT

Environment variable name, which is used to obtain the URL and upload the inference result to the AI platform.

upload_infertime_pushgateway

PUSHGATEWAY_INFERTIME_METRICS_ADDRESS

Environment variable name, which is used to obtain the URL and upload the inference latency to the AI platform.

ai_platform_http_client.ca

None

CA certificate of the server, which is used by the client to verify the server when https is enabled and images are uploaded to the third-party platform.

ai_platform_http_client.crl

None

CRL of the third-party server configured when https is enabled and images are uploaded to the third-party platform.

http_pushgateway_client.ca

None

Client certificate to be configured when the inference result is reported to the third-party Prometheus. It is used by the client to verify the server.

http_pushgateway_client.crl

None

CRL of the third-party Prometheus server configured when the inference result is reported to the third-party Prometheus.

ca_retention_time

None

Expiration time of the certificate retention period, which is automatically set when the certificate is imported. The default retention period is one year.

upload_queue_length

100

Maximum cache length of the queue when the function of uploading images to a third party is enabled.

The value range is [1,1000].

rate_limit

1000/minute

Flow control. The value range is [1,3000].

This parameter is not shared among multiple application processes. If this parameter is set to 1000/minute, the flow control limit of each application process is 1000/minute. Set this parameter based on your requirements.

get_result_timeout

5000

Timeout interval for obtaining the processor inference result.

The value range is [1000,60000].

sdp_algorithm_id

9

Encryption algorithm, which is used when the encryption script is executed and https is enabled.

  • AES128_GCM = 8
  • AES256_GCM = 9

ssl_ciphers

ECDHE-ECDSA-AES256-GCM-SHA384:ECDHE-RSA-AES256-GCM-SHA384:ECDHE-ECDSA-AES128-GCM-SHA256:ECDHE-RSA-AES128-GCM-SHA256:ECDHE-ECDSA-AES256-CCM:ECDHE-ECDSA-AES128-CCM

TLS encryption algorithm suites, which are used when https is enabled.

ai_server_logging_config.log_dir

logs

Path for storing inference logs.

  • The absolute path can be configured and stored only in the /inference directory.
  • The relative path is created in the mxAOIService directory, for example, logs/xxx.

ai_server_logging_config.formatter

'%(levelname)s %(asctime)s [%(filename)s:%(lineno)d] %(message)s'

Inference log format.

ai_server_logging_config.datefmt

'%Y-%m-%d %H:%M:%S'

Date format of an inference log.

ai_server_logging_config.max_bytes

104857600 B (100 MB)

Size of a log file.

The value range is [1e6,1e9].

ai_server_logging_config.backup_count

10

Number of inference log backups.

The value range is [1,100].

colloid_detection_min_area_n

0.3

After the BlobMask and Blob information (ID, label, and area) of LabelMask is calculated, filter out the Blob with an extremely small area. The threshold is defined as n times the minimum Blob area of the template.

The value range is (0,1].

image_size_limit

20971520 B (20 MB)

Size of an inference image. The maximum value is 30 MB.

The value range is [1,31457280].

image_number_limit

10

Upper limit of the image quantity. The number of images sent in a single inference request cannot exceed 10.

The value range is [1,10].

ip_white_list

[]

Trustlist of user device IP.

  • By default, there is no restriction.
  • Set a trustlist, and only the IP addresses in the trustlist can access the system.

app_multi_process_num

5

Number of processes started by the RESTful app.

The value range is [1,10].

max_model_multi_process_num

10

Number of processes for model loading.

The value range is [1,40].

request_max_content_length

31457280 B (30 MB)

Maximum body length allowed in a request. The maximum value is 200 MB.

The value range is [1,209715200].

kmc_config_sem_key

0x20161516

KMC key file lock semaphore. The default permission is 600.

If a certificate is imported by different users, this value needs to be changed. Otherwise, the certificate may fail to be imported. For details about how to rectify the fault, see 4.

The value range is [0x20161112, 0x20169999].

http_keepalive_time

30

Keep-alive time of the connection between the client and the server. The default value is 30 seconds.

The value range is [1,120].

support_scene_list

- detection/DetectionInferScene

- ocr/OCRInferScene

- segmentation/SegmentationInferScene

- detection_detection/DetDetInferScene

- detection_ocr/DetOCRInferScene

- detection_segmentation/DetSegInferScene

- tag_paste/TagPaste

- wafer/WaferInferScene

- detection_htp/DetHTPInferScene

Default scenarios supported by the inference service.

wafer/WaferInferScene is embedded in the semiconductor service software package, and the other seven scenarios are built in the CCC semiconductor service package.

scenes_module_map

det: det_cls

det_cls: det_cls

det_det: det_det

ocr: ocr

seg: det_seg

det_ocr: det_ocr

det_seg: det_seg

tag_paste: tag_paste

wafer: wafer

det_htp: det_htp

Scenario names and labels supported by the inference service.

wafer: wafer is embedded only in the semiconductor service software package. Other scenarios are embedded in the CCC semiconductor service package.

scenes_map

det: detection

det_cls: detection

det_det: detection_detection

ocr: ocr

det_ocr: detection_ocr

seg: segmentation

det_seg: detection_segmentation

tag_paste: tag_paste

wafer: wafer

det_htp: detection_htp

Inference scenario names supported by the inference service.

wafer: wafer is embedded only in the semiconductor service software package. Other scenarios are embedded in the CCC semiconductor service package.