Articles on: Problems solution

Error Handling

General information



Since Albato makes HTTP requests to a third-party system, errors may occur. For example, the authorization data is outdated, fields are filled in incorrectly, an error is received by requesting API, or the server is currently unavailable.

Every time Albato makes a request, the system checks a response. If it has an error sign, an operation is considered to be completed with an error. Errors can occur both in an automation and when creating/updating a connection.

Albato sends email notifications to users in case of errors in their automations.

Albato assigns penalty points that can stop an automation, in case of reaching the highest level of these points.

An automation that catches only errors will be stopped. Any automation with errors can be restarted manually.

Error types



Partner (app) error - this error occurs when Albato makes an outgoing HTTP request to an app, and the app returns an indication that the request is invalid and sends the text of the error.

Timeout error - this error occurs when Albato doesn't receive a response to its request. By default, Albato waits 10 seconds for a response, then it reports an error with text: Request for {{ URL }} timed out. This type of error automatically resends data and assigns penalty points to the automation.

Connection error - this error occurs when Albato cannot establish a connection with a server. For example, a third-party system has failed and its servers are unavailable. In this case, an error text displays: Failed to establish a connection with {{ URL }}

Manual errors resending



You can try to restart each error manually, but all data will be re-transferred. That is, if the step settings have been edited after receiving an error, then at the time of resending, the new settings will already be transferred (e.g., if the field set from the previous step has been changed). This mechanism will work only if the step position has not been changed after receiving the error.

Go to the Automation log and click the resend icon to resend data.




After clicking the button, an attempt to resend the data will be made within 5 minutes. An additional note will appear in the same block of the log below. In this case, if there are next steps after the erroneous step, they will also be processed.

Manual errors restart is unlimited, while the automation is running.

Automatic error resending



Only Timeout Error type can be automatically resent. In this case, Albato considers a server as temporarily unavailable.

If an error with the Timeout Error type occurs, Albato will try to resend this error 5 times with the following intervals:

in 5 minutes
2.in 15 minutes
3.in 30 minutes
4.in 60 minutes
5.in 60 minutes

If timeout continues to persist, attempts are stopped. After the 5th attempt, an email notification will be sent. All automatic resends do not assign penalty points if the timeout is repeated.

Error notifications



If you put this checkbox in your account settings, Albato will send you an email notification every hour in case of automation errors. By default, Albato sends notifications to the email specified during your registration, but you can provide additional email in your account settings. Notification email describes errors, a list of automations in which an error occurred, and their time periods. If the error is of type Timeout Error, the email will not be sent until all 5 automatic resend attempts have been used up.



How the penalty point system works



We mentioned above that each automation has its own set of penalty points. They are assigned to a particular automation. A penalty point will be given for the following types of errors:

Timeout error
Connection error

Each time an error of the specified types occurs in the automation, it is assigned 1 point. In this case, the work of the automation will be slowed down or stopped completely. There are several thresholds for reaching penalty points.

If the automation has an API-trigger, when reaching:

3 points - the pause between trigger runs increases by 10 minutes
10 points - the pause between trigger runs increases by 30 minutes
15 points - the pause between trigger runs increases by 60 minutes
20 points - the automation is finally stopped

If the request in the automation is executed successfully (It doesn't matter if it received data or not - it can be an empty "No new data" request or if the trigger didn't pass the filter), then the threshold is reset exactly 1 step back. That is, if 10 points were assigned, the next trigger run will be executed in 30 minutes, but if the request is successful, the number of points will be reset to 3, and the pause will be 10 minutes, if after 10 minutes there will be a successful request again - the points will be reset to zero.

Automation that has stopped can only be started manually.

If the automation has a webhook trigger, it will not be slowed down, but points will be accumulated. Since it is impossible to track and freeze requests from another system, then if some threshold is reached, instead of increasing the pause, the same amount of time points will simply not be assigned. But if the timeout is kept for a long time - the automation will stop. It makes no difference which step gets an error - a trigger or an action - a point will be assigned in any case.

When automation stops



If the automation explicitly contains only failed requests and no successful ones, it will be stopped after some time. It will only be possible to restart it manually after fixing errors. In this case, the type of error does not matter, all errors in the automation are taken into account.

There is a general system of error checks in automation. All checks are performed exactly at noon, according to the time zone selected in the profile. There are two types of checks.

Once every 24 hours

If there are more than 50 runs in the scenario, and each run has an error, and there is not a single successful cycle (where all steps were executed successfully) - such an automation will be stopped. If there were less than 50 runs or at least one successful run - the automation will not be stopped. If the automation is stopped, a notification letter is sent.

Once every 7 days

If the automation has only failed requests and no successful requests for 7 days, the automation will be stopped. In this case, the number of requests and errors is irrelevant. That is, if the scenario runs only once in a week and gets an error, it will be stopped, but if there is at least one successful request (even one that got no data), the automation will not be stopped. A notification email is sent when the automation is stopped.

Updated on: 06/09/2023

Was this article helpful?

Share your feedback

Cancel

Thank you!