There were few job failures and the UI encountered a brief outage in some API calls on May 3 between 16:07 and 16:14 GMT+2/CEST.
Affected jobs and API calls returned an Application error.
The outage was caused by an automatic upgrade of one of our metadata DB servers. We will be turning off automatic upgrades to prevent this in the future.
We're sorry for this inconvenience.
There were few jobs failures between 16:00 - 17:00 CET and 21:00 - 22:30 CET. We have identified the root causes of the problem and rollbacked to previous version.
Affected jobs were returning error: Authentication token has expired. The user must authenticate again
We're sorry for this inconvenience.
Snowflake data previews in Storage were returned in wrong encoding since this Friday which caused that some characters were missing.
This is fixed now. Stored, imported and exported data wasn't affected by this bug.
Sorry for any inconvenience.
Between Mar 29 10:03 UTC and Mar 29 13:10 UTC all Snowflake Writer jobs had accidentally 15 seconds limit and potentially ended-up by error message:
Statement reached its statement or warehouse timeout of 15 second(s) and was canceled.
tldr; If you're using the old Facebook Ads (Deprecated) extractor then migrate to new extractor or set the `api_version` attribute to v2.8 in corresponding configuration sys table of the old extractor before April 25 2017.
Marketing API(aka Facebook Ads API) version v2.7 has been marked as deprecated and will stop working from April 25, 2017. The latest version is v2.8, see Marketing API Changes in v2.8.
The New Facebook Ads extractor uses version v2.8 by default so no actions are required.
The Old Facebook Ads extractor uses version v2.7 by default. The best thing to do would be to migrate to the new extractor. Another possibility is to override the api version setting: Go to the configuration sys table and update the attribute api_version with value v2.8. There should be no incompatibilities, but we offer no guarantees.
As you might have already noticed, new versions of Facebook Extractor and Facebook Ads Extractor, that runs on container based architecture, are available. The old ones have been marked deprecated.
Due to various differences in output tables structure and configuration between new and old Facebook extractors we don't provide automatic migration tool, however there is basic tutorial on how to create configuration with same authorized account. See here for Facebook and here for Facebook Ads extractor.
We are always glad to receive feedback, so if you have any questions or ideas how to improve this component, please don't hesitate to contact us. Go ahead and try it ;)
We no longer provide Keboola MySQL database provisioning for database writers. Existing Keboola provisioned MySQL databases will be kept intact till the end of Jun 2017. Existing MySQL users will have to move to their own MySQL instances. Here are a few suggestions of vendors which provide MySQL provisioning:
You can keep your data inside KBC and connect to them directly. We believe that our Snowflake Writer is a great replacement for your existing MySQL Writer, so please check it out!
Do not hesitate to contact if you have any questions!
Signing into KBC with your Google account is not working presently.
We're investigating and hope to have it resolved shortly.
In the meantime, please sign in using the standard email and password method.
Sorry for this inconvenience, this post will be updated when further information becomes available.
UPDATE 08:21 UTC We have identified the issue. The fix will be deployed in ASAP.
UPDATE 08:55 UTC Signing in with your Google account is again fully operational.
We've noticed some write jobs to GoodData have been failing since their most recent maintenance period.
We're investigating the issue and will update here as soon as possible.
Update 11:30AM March 12, 2017
The failures were due to a breaking change update to the GoodData date dimension API during their last maintenance period. It should only affect projects which are using custom date dimensions. We are currently assessing options to remedy the situation.
Update 20:30 CET March 12, 2017
For any orchestration that failed to write to GoodData this weekend, in order to have it run correctly, please edit the task and change the action from `upload-project` to `load-data`. It should then work correctly if you re-run it.
Please note however that new writers with custom time dimensions, or changes to old configurations which had them, may still fail.
Update 13:20 CET March 13, 2017
GoodData has fixed the issue. All existing configurations should work now. The switch from `upload-project` to `load-data` is no longer required, but we still recommend it as best practice to prevent possible future issues.
Update 07:00 CET March 14, 2007
In some cases the GoodData writer jobs are still failing. Switching to `load-data` is the correct remedy. Read what
GoodData has to say.
Update 15:47 CET March 14, 2007
The issue seems to have been completely resolved as per GoodData Support.