This is an automated email from the ASF dual-hosted git repository.
michaelsmith pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/impala.git
The following commit(s) were added to refs/heads/master by this push:
new 1bddbefb2 IMPALA-14580: Document Iceberg table repair functionality
1bddbefb2 is described below
commit 1bddbefb2ddf08d52086df31e118399e5de1d2a2
Author: Noemi Pap-Takacs <[email protected]>
AuthorDate: Thu Nov 27 16:36:38 2025 +0100
IMPALA-14580: Document Iceberg table repair functionality
Testing: built docs locally
Change-Id: I67a861a56269648c5f8c2e9697861bf95587f731
Reviewed-on: http://gerrit.cloudera.org:8080/23738
Tested-by: Impala Public Jenkins <[email protected]>
Reviewed-by: Daniel Vanko <[email protected]>
Reviewed-by: Zoltan Borok-Nagy <[email protected]>
---
docs/topics/impala_iceberg.xml | 37 +++++++++++++++++++++++++++++++++++++
1 file changed, 37 insertions(+)
diff --git a/docs/topics/impala_iceberg.xml b/docs/topics/impala_iceberg.xml
index f8e180685..0892f04e0 100644
--- a/docs/topics/impala_iceberg.xml
+++ b/docs/topics/impala_iceberg.xml
@@ -842,6 +842,43 @@ ALTER TABLE ice_tbl EXECUTE remove_orphan_files(now() -
interval 5 days);
</conbody>
</concept>
+ <concept id="iceberg_repair_metadata">
+ <title>Repair table metadata</title>
+ <conbody>
+ <p>
+ Users should always use the engine/Iceberg API to interact with
Iceberg tables;
+ e.g. to remove a partition, use Impala and issue the DROP PARTITION
statement
+ instead of deleting the partition directory.
+ Deleting files directly from storage without going through the Iceberg
API
+ corrupts the table, and makes queries that try to read the missing
files fail
+ with the following error message:
+ <codeph>Iceberg table [...] cannot be fully loaded due to unavailable
+ files</codeph>.
+ </p>
+ <p>
+ This happens because the metadata files are still referencing the
missing data
+ files. This erroneous state can be fixed by restoring the deleted
files on the
+ file system.
+ If this is not intended or not possible, the dangling references can
be removed
+ from the Iceberg metadata with the
+ <codeph>ALTER TABLE ... EXECUTE repair_metadata()</codeph>
+ statement, so that the table becomes functional again.
+ <codeblock>
+-- Use the statement simply without parameters:
+ALTER TABLE ice_tbl EXECUTE repair_metadata();
+ </codeblock>
+ </p>
+ <note>
+ This operation does not restore the deleted content. Execute only if
+ there is no intention to restore the missing data.
+ <p>
+ Impala can repair the table only if the missing files are data files,
+ but it cannot repair the table if there are missing delete files.
+ </p>
+ </note>
+ </conbody>
+ </concept>
+
<concept id="iceberg_metadata_tables">
<title>Iceberg metadata tables</title>
<conbody>
