Around 1:30am CEST one of the job worker instances stopped processing assigned jobs. This could have lead to jobs being stuck in the processing state for a long time without any activity.
At 11:15am CEST the worker instance was terminated and all unfinished jobs started processing on other instances.
We're sorry for this inconvenience.
The maintenance period in the US region has just finished. All systems are back online and working properly.
Thank you for your patience.
The maintenance period in the EU region has just finished. All systems are back online and working properly.
Thank you for your patience.
The pre-announced maintenance period for the US and EU regions has just started. The estimated downtime is 5 hours.
There will be a maintenance period on Saturday, October 6th, 2018 from 8:00am CEST and should take less than 5 hours.
We will be upgrading component job indices and metadata databases.
All projects in connection.keboola.com and connection.eu-central-1.keboola.com will be inaccessible during the maintenance.
Since September 25 we're experiencing degraded Snowflake performance affecting all Snowflake operations.
We're sorry for this inconvenience, we're working with Snowflake to fix this issue.
Update, October 8
Snowflake Engineering team has discovered and fixed the issue (waiting for an official statement from Snowflake). We're seeing operation times going back to normal.
We have encountered an increased error rate in MySQL Extractor after a new version release on Friday, Sept 14, 16:00 UTC.
These errors happened only when the job had a connection issue and had to reconnect to the MySQL server.
The previous version is now rolled back and we're working on a fix.
We're sorry for this inconvenience.
We have encountered an increased number of Snowflake connection errors between 04:11 UTC and 04:21 UTC. This may have caused failed storage and component jobs.
Furthermore Snowflake announced possible SQL query failures between 02:24 UTC and 03:30 UTC.
We're sorry for this inconvenience, all systems are operational now.
We're currently investigating an incident in the US region that may have caused component job and orchestration failures today, Tuesday, Sep 11, between 02:30 UTC and 06:30 UTC.
Update 08:30 UTC
The DB storing job locks (preventing jobs from running multiple times on multiple workers) was restarted at 02:33 UTC. All connections were terminated and all component jobs and transformations running at that time were disconnected. This could lead to any or both of the following situations:
Any jobs started after the DB restart were not affected by this issue.
We apologize for this inconvenience. We're planning an infrastructure change to prevent such huge impact during similar situations.
If you have any further questions, please use the support button in your project.
Between 4:00 pm CEST and 7:30 pm CEST the new version of Create Manifest processor was failing with an application exception. This change was rolled back and the processor is functional now.
We're sorry for this inconvenience.