search

asbassan / pulserepo

0 stars 0 forks source link

Databricks #9

Open asbassan opened 3 years ago

asbassan commented 3 years ago

Support ticket took very long time to resolve. limited error reporting there. it tooks weeks of set up followed by days of sevA calls. Azure networking and Databricks had to get together to identify the problem for us. Dissatisfied with pace and timings of the issue resolution.

asbassan commented 3 years ago

I need to look at the ticket and then get the RCA on steps. Any MIP that could help them. Team up with Henry on troubleshooting workshop.

asbassan commented 3 years ago

Problem : Lack of detailed error logging during Azure Databricks set up. Description: There is a lack of error logging for Azure Databricks service related to the cluster set up. While the error message does point to the possible error, it does not provide actionable insights. This leads to the customer opening a support ticket and many instances of log collection before they can pinpoint to the exact settings. An example of the error message is: “Spark failed to start: Could not connect to driver instance. Possible reason: network misconfiguration”. Support ticket 2105140040004874 has the context. There is a difference in the reverse IP address lookup response time between East US and West US. These difference in the response time causes the reverse lookup to fail and hence some services fail to set up. Ask: 1) Include the details of cause of the error message rather than a generic error message. 2) Perform prerequisite network configuration tests before setting up the service so that the customer can correct the network configuration before set up begins. 3) Perform prerequisite DNS iP resolver tests before setting up the service so that the customer can correct the network configuration before set up begins

asbassan commented 3 years ago

Having better data in the logs.