Edge Data Replication Fails with High-Precision Timestamps

[KaarelRaeis-SOI]

Description:
I'm experiencing issues with InfluxDB replication from Influx OSS 2.7.11 on GC vm with plenty of resources when data points have timestamps with high precision (nanoseconds). Each edge sensor(100+ sensors, up to tens of measurements per sensor) reading has a unique timestamp with many decimal places. Replication fails under these conditions after just a couple of seconds of data getting pushed to Influx Cloud also on GC. Remaining bytes to be synced keeps increasing but nothing is replicated despite latest 204 status.

However, when I resample the data to have timestamps without decimals (e.g., whole seconds) via influx task, replication proceeds without issues, data volume is unchanged.

Influx task

option task = {name: "resample1s", every: 1s, offset: 1s}

from(bucket: "openrvdas")
    |> range(start: -5s)
    // Adjust as needed
    |> last()
    // Get only the most recent value for each series
    |> to(bucket: "resampled1s")

Steps to Reproduce:

1. Write data points with nanosecond precision timestamps to the source InfluxDB instance.
2. Set up replication to a target InfluxDB instance.
3. Observe that replication fails.
4. Resample the data to have integer second precision timestamps via influx task to separate bucket.
5. Observe that replication succeeds.

Expected Behavior:
Replication should handle high-precision timestamps without failing.

Actual Behavior:
Replication fails when data points have high-precision timestamps but succeeds when timestamps are resampled to whole seconds.

Environment:

• InfluxDB Version: 2.7.11
• Operating System: Ubuntu 22
• Deployment: GC vm

Logs:
[Include any error messages or logs that could help diagnose the issue.]