[BUG] Timestamp mismatch

[rwejdling]

Describe the bug
Timestamp in incident updates aren't matching the configured timezone.

Version
1.10.1 and previous versions (just never got around to reporting it)

To Reproduce
Steps to reproduce the behavior:

1. Set timezone: CET or timezone: Europe/Copenhagen in config.yaml
2. Post an update on a incident
3. Validate that the Posted At: timestamp on the update in digest channel is actually correct

Expected behavior
Set posted at to the correct timestamp

Screenshots

Environment (please complete the following information):

• Platform: container running on ECS

Additional context
N/A

[rwejdling]

Actually all timestamps are one hour off:

I'm currently in Copenhagen which is CET (GMT+1).

[uptickmetachu]

I believe this might be because all the timestamps are currently stored as strings with the timestamp embedded in them:

updated_at is 2024-02-13T20:53:20 AEDT.

Would it be feasible to refactor the time columns from string/varchar to utc timestamps and then render them to the users specified timezone? @echoboomer

[uptickmetachu]

application_timezone = 'Europe/Copenhagen'


def fetch_timestamp(short: bool = False):
    """Return a localized, formatted timestamp using datetime.now()"""
    now = datetime.now()
    localized = timezone(application_timezone).localize(now)
    if short:
        return localized.strftime(timestamp_fmt_short)
    return localized.strftime(timestamp_fmt)

This seems to be the problem.

incident-bot/backend/bot/shared/tools.py

Line 22 in ec7b116

def fetch_timestamp(short: bool = False):

This should be

    now = datetime.datetime.now(datetime.UTC)
    localized = now.astimezone(timezone(application_timezone))
    if short:
        return localized.strftime(timestamp_fmt_short)
    return localized.strftime(timestamp_fmt)