Checking failed jobs
When working with cliques in YTsaurus, failures sometimes occur — for example, queries complete with errors or are interrupted. These failures cause jobs to fail within a YT operation. To quickly identify and resolve the issue, you need to be able to analyze the logs of failed jobs.
Warning
If the number of failed jobs in an operation exceeds 100, logging will stop — the logs won’t be saved. Therefore, it’s important to respond to failures promptly to avoid losing information.
How to view and analyze logs
-
Open the YT operation web interface (1), as described in the section How to open the YT operation web interface.
-
On the right, in the Failed block (2), check the number of failed jobs.
-
Go to the Jobs tab (3) in the tab bar.
-
In the job list that opens, use the
State:Failedfilter (4) to find the failed jobs. -
In the row with the failed job, find the Error / Debug column.
-
To download the job log, click
in the Stderr field (5) . -
Review the log and pay attention to the following:
- error messages (usually contain the keywords Error, Exception, Failed);
- timestamps (to understand exactly when the failure occurred);
- mentions of resources (for example, tables or files) that the job was working with.
Note
Save the logs of failed jobs — they will be useful for:
- monitoring system stability;
- analyzing recurring errors;
- providing data to support teams.