LOC’s book sorter machine has an issue with data lag/sync between FOLIO and the machine.

• Book sorter machine integrates with Folio via SIP2

• The process involves 2 calls to Folio

  • 1. The item barcode is scanned, and a call is made to identify the item.

  • 2. Using the returned response, the item’s location is established, and the second call is made to check in the item to the appropriate tenant.

• The book sorter machine will then deliver the item to the appropriate bin for delivery to be shelved.

• Check-in rate corresponds to a worker manually placing each book on the book sorter machine (every few seconds at peak)

• The book sorter will time out during the operation, fail to sort the book, and deliver the book to the error bin instead.

  • The book sorter machine has a hard-coded timeout of 1 second for the entire process

  • FOLIO will respond to each call within approximately 2 seconds, with an overall response time of 4 seconds or more.

• Each executed command performs a login request with the provided pair of credentials.

• Credentials are provided during the login command (username | password | servicePointId).

• Each new pair of credentials could affect the authorization cache in folio-module-sidecar (uses sessionId for authorization cache-key)

• Token cache was disabled in the following stories:

  • https://folio-org.atlassian.net/browse/SIP2-222

  • https://folio-org.atlassian.net/browse/SIP2-224

• Okapi request caching for authorized requests:

  • ProxyService.java:checkTokenCache

• Eureka Sidecar caching for authorized requests:

  • KeycloakAuthorizationFilter.java

The total number of requests per the item sorting series of requests is 10:

• 2 for the Login command

• 5 for the Item Information command

• 3 for the Item Checkin command

Although each login operation is relatively quick, performing it 10 times within a single session significantly impacts performance. To optimize this, an access token should be issued once at the beginning of the session and refreshed as needed using the HTTP POST /authn/refresh Token API request. If the refresh operation fails for any reason, a new access token should be obtained and stored in the cache as a fallback.

edge-sip2 container being used on Volaris Rancher environment with increased memory limits (with default ~120Mi container is unstable):