anuragmantri commented on PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#issuecomment-2686319047
Thanks for attempting to fix this @karuppayya. What happens if the `data`
and `metadata` locations evolve (if that's possible)?
At some point, users may decide to change the met
karuppayya commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1972303374
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -301,24 +303,34 @@ private Dataset actualFileIdentDS()
ismailsimsek commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1960165781
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -335,6 +347,21 @@ private Dataset listedFileDS() {
ismailsimsek commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1960165781
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -335,6 +347,21 @@ private Dataset listedFileDS() {
dramaticlly commented on PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#issuecomment-2664203151
> How about introducing the UUID in the table path while creating table?
This way two tables cannot share a same location.
I think that's what trino did with `iceberg.unique-t
dramaticlly commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1958833411
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -84,11 +86,11 @@
* comparing the actual files in tha
dramaticlly commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1958833411
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -84,11 +86,11 @@
* comparing the actual files in tha
ajantha-bhat commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1958839422
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -301,24 +303,34 @@ private Dataset actualFileIdentDS(
karuppayya commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1958822862
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -335,6 +347,21 @@ private Dataset listedFileDS() {
Fokko commented on code in PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#discussion_r1957869157
##
spark/v3.5/spark/src/main/java/org/apache/iceberg/spark/actions/DeleteOrphanFilesSparkAction.java:
##
@@ -335,6 +347,21 @@ private Dataset listedFileDS() {
retu
szehon-ho commented on PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#issuecomment-2662233970
I think it makes some sense to me, but maybe better to put it behind a flag
as its a behavior change? Something like -- listCurrentPaths? Curious what
others think?
--
This is
karuppayya commented on PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#issuecomment-2660448261
cc: @aokolnychyi @RussellSpitzer @szehon-ho @anuragmantri @dramaticlly
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to
karuppayya commented on PR #12278:
URL: https://github.com/apache/iceberg/pull/12278#issuecomment-2660441981
This change doesnt solve the following cases:
1. Tables that the same location root
2. Tables that are that have root as data or metadata directories of a
different table
karuppayya opened a new pull request, #12278:
URL: https://github.com/apache/iceberg/pull/12278
### Issue
Tables t1 and t2 use a common prefix for their table location
`/path/to/shared_root`.
t1 has it data and metadata dir -> `/path/to/shared_root/t1`
t2 has its data a and metad
14 matches
Mail list logo
