[jira] [Created] (IMPALA-12124) Add a way to get the current file from a minidump for an IO / scanner thread

["Joe McDonnell (Jira)" <ji...@apache.org>]

Joe McDonnell created IMPALA-12124:
--------------------------------------

             Summary: Add a way to get the current file from a minidump for an IO / scanner thread
                 Key: IMPALA-12124
                 URL: https://issues.apache.org/jira/browse/IMPALA-12124
             Project: IMPALA
          Issue Type: Improvement
          Components: Backend
    Affects Versions: Impala 4.3.0
            Reporter: Joe McDonnell


In the past, there have been bugs in how Impala handles corrupt data files. This can be difficult to diagnose without knowing the name of the corrupt data file, but this is not easy to find if Impala crashes.

Impala has some thread information that it keeps on the stack to help identify threads and the current query if Impala crashes. These are printed to the Impala log by the minidump logic, and it can be retrieved using gdb after converting the minidump to a core.

It would be useful to have something similar for the file / scan range that the DiskIOMgr thread or scanner thread is working on. This would speed diagnosis for these corruption cases (and can identify the table that is being scanned).



--
This message was sent by Atlassian Jira
(v8.20.10#820010)