BerriAI/litellm · error · ValueError

Could not parse timestamp '{timestamp_str}': {e}

Error message

Could not parse timestamp '{timestamp_str}': {e}

What it means

ValueError raised by CloudZeroStreamer._parse_and_convert_timestamp when datetime.fromisoformat rejects the input string after the Z-suffix and explicit-offset handling. The helper only understands ISO 8601; anything else — slash dates, epoch numbers, locale-formatted strings, empty strings, or precision variants your Python version can't parse (pre-3.11 fromisoformat is strict) — raises with the underlying parser message appended.

Source

Thrown at litellm/integrations/cloudzero/cz_stream_api.py:149

                    "+08:00",
                    "+09:00",
                    "+10:00",
                    "+11:00",
                    "+12:00",
                )
            ):
                dt = datetime.fromisoformat(timestamp_str)
            else:
                # Assume user timezone if no timezone info
                dt = datetime.fromisoformat(timestamp_str)
                if dt.tzinfo is None:
                    dt = dt.replace(tzinfo=self.user_timezone)

            # Convert to UTC
            return dt.astimezone(timezone.utc)

        except ValueError as e:
            raise ValueError(f"Could not parse timestamp '{timestamp_str}': {e}")

    def _send_daily_batch(self, batch_date: str, batch_data: pl.DataFrame, operation: str) -> None:
        """Send a single daily batch to CloudZero API."""
        if batch_data.is_empty():
            return

        headers: Final = {
            "Authorization": f"Bearer {self.api_key}",
            "Content-Type": "application/json",
        }

        # Use the correct API endpoint format from documentation
        url: Final = f"{self.base_url}/v2/connections/billing/anycost/{self.connection_id}/billing_drops"

        # Prepare the batch payload according to AnyCost API format
        payload: Final = self._prepare_batch_payload(batch_date, batch_data, operation)

        try:

View on GitHub (pinned to 77b7c6c40c)

Solutions

1. Normalize timestamps to ISO 8601 before export: dt.isoformat() or keep 'YYYY-MM-DDTHH:MM:SS(.ffffff)(Z|±HH:MM)'
2. If values are epoch seconds/millis, convert first: datetime.fromtimestamp(int(ts), tz=timezone.utc).isoformat()
3. Coerce empty/NULL to a sensible default (row creation time) instead of passing ''
4. Pin Python >= 3.11 in the image if inputs carry less-common ISO variants

Example fix

# before
streamer.stream_usage_data(df.with_columns(pl.col("startTime").str.to_str()))  # raw '2024/07/15 00:00:00'

# after
df = df.with_columns(
    pl.col("startTime")
      .str.strptime(pl.Datetime, "%Y/%m/%d %H:%M:%S", strict=True)
      .dt.replace_time_zone("UTC")
      .dt.to_string("%Y-%m-%dT%H:%M:%S%z")
)

from datetime import datetime

def is_parseable_timestamp(ts: str) -> bool:
    if not isinstance(ts, str) or not ts.strip():
        return False
    try:
        datetime.fromisoformat(ts.strip().replace("Z", "+00:00"))
        return True
    except ValueError:
        return False

try:
    dt_utc = streamer._parse_and_convert_timestamp(ts)
except ValueError as e:
    # convert non-ISO inputs yourself, e.g. epoch seconds:
    from datetime import datetime, timezone
    dt_utc = datetime.fromtimestamp(int(ts), tz=timezone.utc)

When it happens

Trigger: Feeding the streamer usage rows whose timestamp column is '2024/07/15 00:00:00' (slashes), '1699999999' (epoch as string), '15-07-2024' (day-first), '' (empty), or a nanosecond-precision value on Python < 3.11. Note that naive strings are assumed to be in the configured user_timezone, so missing offsets alone do not raise.

Common situations: Usage data extracted from databases with non-ISO timestamp formatting (e.g. Snowflake/BigQuery variants rendered by drivers); spreadsheets/CSVs with locale dates; Python version differences between dev (3.11+) and prod (3.10) containers; columns accidentally containing NULLs coerced to ''.

Related errors

• CloudZero configuration missing. Please set CLOUDZERO_API_KE
• Generated CZRN is invalid: {czrn}
• Invalid CZRN format: {czrn}
• Database not connected. Connect a database to your proxy - h
• limit must be an integer

AI-assisted analysis of BerriAI/litellm@77b7c6c40c (2026-08-18). Data as JSON: /api/errors/b8ad79bf7cca79df. Report an issue: GitHub.