问题
I have a BigQuery dataset of Gmail logs which is managed by Google. Google streams the email logs to a table named daily_ near real time which is partitioned by day. In the logs are records indicating when certain messages trigger mail routing rules, and I want to select those and calculate the amount of time required to execute the routing rule with a time diff. I was successful in calculating the time diff using SQL, saved it as a view, and now want to graph it in DataStudio. However, whenever I add a time filter with a date range dimension, it gives me the following error:
User Configuration Error
This data source was improperly configured.
The query returned an error.
Invalid timestamp: '--' Error ID: 3d446739
My first thought was that perhaps there are rows where the partition _TABLE_SUFFIX field is blank or null. However, after running a "select distinct _TABLE_SUFFIX" and reviewing the results I'm no longer convinced that is the case, they all have a date value.
Any ideas what might be causing this? For what it's worth, when I add the view as a datasource to DataStudio, it correctly selects the datatype as a date. Google stores values into _TABLE_SUFFIX with the formal YYYYMMDD which is exactly how DataStudio sees it when adding the source.
The worst part about all of this is that a colleague of mine is using the same dataset as a source (albeit with a different but similarly configured view) and his works fine while mine doesn't. Any help is appreciated.
This is the SQL view that I created and want to graph in Data Studio:
SELECT
_TABLE_SUFFIX as pt,
MIN(DATETIME(timestamp_micros(event_info.timestamp_usec), "America/New_York")) as time_before,
MAX(DATETIME(timestamp_micros(event_info.timestamp_usec), "America/New_York")) as time_after,
DATETIME_DIFF(
MAX(DATETIME(timestamp_micros(event_info.timestamp_usec), "America/New_York")),
MIN(DATETIME(timestamp_micros(event_info.timestamp_usec), "America/New_York")),
SECOND) as timediff,
message_info.num_message_attachments,
message_info.source.address as sender,
dest.address as recipeint,
message_info.rfc2822_message_id,
message_info.subject
FROM
`g-suite-logs.gmail_logs.daily_*` as t1,
UNNEST ( message_info.destination ) as dest,
UNNEST ( message_info.triggered_rule_info ) as rule
WHERE rule.rule_name = "Route to third party MTA" OR rule.rule_name = "Receive back from third party MTA"
GROUP BY
pt,
message_info.rfc2822_message_id,
message_info.source.address,
dest.address,
message_info.subject,
message_info.num_message_attachments
ORDER BY pt,rfc2822_message_id
I had also suspected that perhaps the UNNEST was introducing some null values to _TABLE_SUFFIX but no longer suspect that is the case.
来源：https://stackoverflow.com/questions/56581720/the-query-returned-an-error-invalid-timestamp