Custom backoff in retryDelayOptions doesn't change the 300ms delay

[aaronatbissell]

The customBackoff function that can be passed to retryDelayOptions doesn't appear to change the retry delay at all. The delay still appears to be 300ms no matter what. The message printed on screen also indicates 300ms no matter what.

Here's the client we are using:

this.client = contentstack.client({
      retryLimit: 4,
      timeout: 60 * this.SEC_TO_MS,
      /**
       * Exponential backoff logic:
       * retry  |  total elapsed time since original request (sec)
       * 0	    |  0.000
       * 1	    |  0.250
       * 2	    |  0.750
       * 3	    |  1.750
       * 4	    |  3.750
       */
      retryDelayOptions: {
        customBackoff: (retryCount, _error) =>
          0.25 * Math.pow(2, retryCount) * this.SEC_TO_MS,
      },
    });

[aaronatbissell]

Even if you make it something simple like this, it still only waits for 300ms

this.client = contentstack.client({
      retryLimit: 4,
      timeout: 60 * this.SEC_TO_MS,
      retryDelayOptions: {
        customBackoff: (retryCount, _error) => 1000 * retryCount,
      },
    });

[ishaileshmishra]

thanks for raising the issue, We will look into it @aaronatbissell.

[sunil-lakshman]

@aaronatbissell
The main difference between the two code snippets is how the customBackoff function calculates the retry delay.

this.client = contentstack.client({
  retryLimit: 4,
  timeout: 60 * this.SEC_TO_MS,
  /**
   * Exponential backoff logic:
   * retry  |  total elapsed time since original request (sec)
   * 0      |  0.000
   * 1      |  0.250
   * 2      |  0.750
   * 3      |  1.750
   * 4      |  3.750
   */
  retryDelayOptions: {
    customBackoff: (retryCount, _error) =>
      0.25 * Math.pow(2, retryCount) * this.SEC_TO_MS,
  },
});

Custom Backoff Calculation:

• Uses exponential backoff logic.
• Delay is calculated as 0.25 * Math.pow(2, retryCount) * this.SEC_TO_MS.
• For each retry, the delay increases exponentially.
• Example delays:
  • Retry 0:
    • 0.25 * Math.pow(2, 0) * 1000 = 250 ms
  • Retry 1:
    • 0.25 * Math.pow(2, 1) * 1000 = 500 ms
  • Retry 2:
    • 0.25 * Math.pow(2, 2) * 1000 = 1000 ms
  • Retry 3:
    • 0.25 * Math.pow(2, 3) * 1000 = 2000 ms
  • Retry 4:
    • 0.25 * Math.pow(2, 4) * 1000 = 4000 ms

this.client = contentstack.client({
  retryLimit: 4,
  timeout: 60 * this.SEC_TO_MS,
  retryDelayOptions: {
    customBackoff: (retryCount, _error) => 1000 * retryCount,
  },
});

Custom Backoff Calculation:

• Uses linear backoff logic.
• Delay is calculated as 1000 * retryCount.
• For each retry, the delay increases linearly.
• Example delays:
  • Retry 0:
    • 1000 * 0 = 0 ms
  • Retry 1:
    • 1000 * 1 = 1000 ms
  • Retry 2:
    • 1000 * 2 = 2000 ms
  • Retry 3:
    • 1000 * 3 = 3000 ms
  • Retry 4:
    • 1000 * 4 = 4000 ms

Summary:

• First Snippet:

  • Exponential backoff: Delays increase exponentially with each retry.
  • Typically used to more quickly mitigate transient errors by progressively increasing delay.

• Second Snippet:

  • Linear backoff: Delays increase linearly with each retry.
  • Simpler, and delays increase at a constant rate.

Depending on your needs, you might prefer one method over the other. Exponential backoff is generally more effective at handling temporary server overloads, while linear backoff is simpler and may be suitable for other types of retries.

[aaronatbissell]

I understand the difference between exponential and linear back-off.

The problem was that no matter what you put as the customBackoff function, it will always wait for 300ms (and the statement that gets printed to the screen always says "waiting 300ms")

Note: this issue is over a year old and I haven't tested to see if this is STILL an issue

Just tested - this still appears to be an issue with v1.17.0

[aaronatbissell]

Here's a repo that reproduces the issue: https://github.com/aaronatbissell/contentstack-management-javascript-issue-55

[cs-raj]

Hi @aaronatbissell we will be looking into the issue as earliest as possible and we have taken it in the upcoming sprint. We will post the reply once fixed
cc: @harshithad0703 @sunil-lakshman @netrajpatel @shafeeqd959

[vkalta]

Hi @aaronatbissell , apologies for inconvenience. I have found a fix for this issue and the PR can be found here https://contentstack.atlassian.net/browse/DX-1409.
Its currently being reviewed.
Thanks for your patience.