Getting ValueError: time data doesn't match format "%Y-%m-%d %H:%M:%S.%f%z" error

I am trying to subtract start_time_ns from end_time_ns in the pandas data frame by using:
df['time'] = pd.to_datetime(df['end_time_ns']) - pd.to_datetime(df['start_time_ns']) which are given in nanoseconds.
I am reading the csv as pd.read_csv(filename,parse_dates=[2, 3],chunksize=chunksize) where column 2 and 3 are start_time_ns and end_time_ns respectively.
The subtraction works fine for the first chunk, but getting error when applying on 30~GB CSV file. The error I get is :

Traceback (most recent call last):
  File "2rg.py", line 17, in <module>
    df['time'] = pd.to_datetime(df['end_time_ns']) - pd.to_datetime(df['start_time_ns'])
  File "/home/nnazarov/.local/lib/python3.8/site-packages/pandas/core/tools/datetimes.py", line 1050, in to_datetime
    values = convert_listlike(arg._values, format)
  File "/home/nnazarov/.local/lib/python3.8/site-packages/pandas/core/tools/datetimes.py", line 453, in _convert_listlike_datetimes
    return _array_strptime_with_fallback(arg, name, utc, format, exact, errors)
  File "/home/nnazarov/.local/lib/python3.8/site-packages/pandas/core/tools/datetimes.py", line 484, in _array_strptime_with_fallback
    result, timezones = array_strptime(arg, fmt, exact=exact, errors=errors, utc=utc)
  File "pandas/_libs/tslibs/strptime.pyx", line 530, in pandas._libs.tslibs.strptime.array_strptime
  File "pandas/_libs/tslibs/strptime.pyx", line 351, in pandas._libs.tslibs.strptime.array_strptime
ValueError: time data "2023-06-20 20:41:11+00:00" doesn't match format "%Y-%m-%d %H:%M:%S.%f%z", at position 816780. You might want to try:
    - passing `format` if your strings have a consistent format;
    - passing `format='ISO8601'` if your strings are all ISO8601 but not necessarily in exactly the same format;
    - passing `format='mixed'`, and the format will be inferred for each element individually. You might want to use `dayfirst` alongside this.

I am pasting line 816780 as well as an information :

NGN,NGN,2023-06-20 20:30:08.305255+00:00,2023-06-20 20:41:08.317472+00:00,131.243.51.211,144.195.208.70,49851,8801,active,UDP,503,0.000107876,"[0, 52, 46, 405, 0, 0, 0, 0]"
NGN,NGN,2023-06-20 20:40:53.903338+00:00,2023-06-20 20:41:11+00:00,2001:400:0:40::200:205,2001:400:211:81::d1,161,56640,idle,UDP,503,0.0001688016,"[0, 0, 0, 503, 0, 0, 0, 0]"
NGN,NGN,2023-06-20 20:40:53.890268+00:00,2023-06-20 20:41:10.986850+00:00,2001:400:211:81::d1,2001:400:0:40::200:205,56640,161,idle,UDP,503,4.6164e-05,"[0, 503, 0, 0, 0, 0, 0, 0]"

How can I resolve the issue?

IIUC, you have mixed datetimes; with and without the UTC offset specified. [mre]:

import pandas as pd
print(pd.to_datetime(["2023-06-20 20:41:11+00:00", 
                      "2023-06-20 20:41:11",
                      "2023-06-20 20:41:11.890268+00:00"]))

errors with

ValueError: time data "2023-06-20 20:41:11" doesn't match format "%Y-%m-%d %H:%M:%S%z", at position 1. You might want to try:
    - passing `format` if your strings have a consistent format;
    - passing `format='ISO8601'` if your strings are all ISO8601 but not necessarily in exactly the same format;
    - passing `format='mixed'`, and the format will be inferred for each element individually. You might want to use `dayfirst` alongside this.

With pandas v2, you could use a combination of keywords utc=True and format="ISO8601" to avoid the error:

print(pd.__version__)
# 2.0.3

print(
      pd.to_datetime(["2023-06-20 20:41:11+00:00", 
                      "2023-06-20 20:41:11",
                      "2023-06-20 20:41:11.890268+00:00"],
                     format="ISO8601", utc=True)
)

DatetimeIndex([       '2023-06-20 20:41:11+00:00',
                      '2023-06-20 20:41:11+00:00',
               '2023-06-20 20:41:11.890268+00:00'],
              dtype='datetime64[ns, UTC]', freq=None)