Hi, in our dbt project we often have accepted valu...
# ask-elementary-ai
w
Hi, in our dbt project we often have accepted values tests with large lists of accepted values. This seems to be an issue for the
edr monitor
cli command when the tests fail, as it concatenates all of the accepted values into the identifier for the alert. Rather than changing all of these tests to custom sql tests, I was wondering whether there is a way to make the command truncate long alert identifiers?
k
The knowledge sources do not contain enough information to answer your specific question about truncating long alert identifiers caused by large
accepted_values
test parameter lists in
edr monitor
. The knowledge sources do document a related (but distinct) issue — where payload size sent to Microsoft Teams becomes too large when test row samples are included in alerts [bug: Teams payload too large]. However, that issue is about sample data size, not about alert identifier length from accepted values lists. There is no mention in the knowledge sources of: • A flag or configuration option to truncate alert identifiers • Any known workaround specifically for long identifiers caused by large
accepted_values
lists What I can suggest based on the knowledge sources: • You could reduce the verbosity of alert content using the
alert_fields
meta property to limit which fields are shown in alerts [alerts configuration], but this controls displayed fields, not identifiers. • You could consider opening a GitHub issue on the Elementary repository to request this feature, similar to how the Teams payload size issue was raised. For a definitive answer, I'd recommend: 1. Checking the Elementary GitHub issues for any existing reports matching your scenario. 2. Reaching out to the Elementary community on Slack for guidance from the team directly. 👍 0 button 👎 0 button ℹ️ Tip: Mention @elementary.ai in the thread for follow-ups or DM the app for 1-to-1 help.
w
I've bumped this issue, is it worth me also raising in #C02CTC89LAX? @kapa.ai
k
It looks like your message came through as "^^" — could you clarify what you meant? Are you confirming the previous answer, or did you have a follow-up question about Elementary Data? I'm happy to help! 👍 0 button 👎 0 button ℹ️ Tip: Mention @elementary.ai in the thread for follow-ups or DM the app for 1-to-1 help.
Good move bumping the GitHub issue! Regarding posting in the Slack channel (C02CTC89LAX): It's worth knowing that as of mid-2024, Elementary updated their support approach — the Slack channel is now primarily community-maintained, and the team has shifted to prioritising GitHub issues for bug reports and feature requests. [support update] So in your case, since you've already bumped the GitHub issue, that is actually the higher-priority channel for getting the team's attention. Posting in Slack may get community input, but the Elementary team is less likely to respond there than on GitHub. [community & support] That said, a Slack post could still be useful to: • See if other community members have found workarounds • Gauge how widespread the issue is, which can help prioritise the GitHub issue 👍 0 button 👎 0 button ℹ️ Tip: Mention @elementary.ai in the thread for follow-ups or DM the app for 1-to-1 help.
w
Is there a way I can turn off the alert? Using perhaps
skip_alert
or
alert_on_failure
config options, or something similar? @kapa.ai
k
The knowledge sources do not explicitly mention
skip_alert
or
alert_on_failure
config options. However, there are a few relevant options that could help: Mute Alerts (Elementary Cloud) If you're on Elementary Cloud, you can mute a test from the test configuration tab to run tests without triggering alerts. [alerts overview] For OSS, the closest options are: 1. Suppression interval — "snooze" alerts for a configured number of hoursmodels - name: my_model_name meta: alert_suppression_interval: 24 # hoursOr via CLI:
edr monitor --suppression-interval 24
[alerts configuration] 2. Disable alerts globally via
dbt_project.yml
varsvars disable_test_alerts: true disable_warn_alerts: true 3. Filter alerts using CLI selectors to exclude specific tests, tags, or models:edr monitor --filters tags:critical[alerts configuration] None of these are a per-test
skip_alert
style toggle for OSS. If that's what you need, it may be worth adding it as a feature request on the Elementary GitHub. 👍 0 button 👎 0 button ℹ️ Tip: Mention @elementary.ai in the thread for follow-ups or DM the app for 1-to-1 help.
w
So what would you recommend? The only solution I'm seeing is to perhaps remove the test which isn't really a great option. We're using the OSS version. Another suggestion would be perhaps to move the test from the yaml to sql? Would that avoid issues with the payload being too large (I assume the entire test sql isn't added to the payload)? @kapa.ai
k
The knowledge sources don't contain enough detail to confirm definitively whether moving to a custom SQL test would avoid the payload size issue. However, based on what is available: What we know • The large payload issue occurs because test parameters (like the long
accepted_values
list) get concatenated into the alert identifier/payload. [GitHub issue] • A community member noted a similar workaround suggestion — moving to custom SQL tests — in the same GitHub issue thread. [GitHub issue] My recommendations (OSS) 1. Move to a custom SQL test — This is likely your best current option. A custom SQL test wouldn't have a large
test_params
block with hundreds of accepted values concatenated into the identifier, which is the root cause of the problem. This seems like a reasonable workaround even if not ideal. 2. Limit test row samples — You can reduce the payload size by lowering the number of sample rows included in alertsvars test_sample_row_count: 0 # disables sample rows entirelyThis won't fix the identifier length issue but may help with overall payload size. 3. Use
alert_fields
to strip down alert content to the bare minimum while the underlying issue is being addressed. I'd suggest commenting your specific use case (the identifier length problem from
accepted_values
) on the existing GitHub issue to help the Elementary team understand the full scope of the problem. 👍 0 button 👎 0 button ℹ️ Tip: Mention @elementary.ai in the thread for follow-ups or DM the app for 1-to-1 help.