GitHub user NicoK opened a pull request:

https://github.com/apache/flink/pull/4024

FLINK-6782[docs] update snapshot documentation to reflect flink 1.3

please also apply to master after accepting

You can merge this pull request into a Git repository by running:

$ git pull https://github.com/NicoK/flink flink-6782

Alternatively you can review and apply these changes as the patch at:

https://github.com/apache/flink/pull/4024.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

This closes #4024

commit 847b87d2fa3e8f0ec4dc4371bc6cffbc3f922925
Author: Nico Kruber <nico@data-artisans.com>
Date: 2017-05-31T08:50:39Z

[hotfix][docs] update Checkpoint docs with correct code example

commit 7554da3d546749d0a5aa6d31d36c03cc8556cc80
Author: Nico Kruber <nico@data-artisans.com>
Date: 2017-05-31T11:51:12Z

FLINK-6782[docs] update snapshot documentation to reflect flink 1.3

Github user alpinegizmo commented on a diff in the pull request:

https://github.com/apache/flink/pull/4024#discussion_r119358154

— Diff: docs/setup/savepoints.md —
@@ -129,11 +144,11 @@ $ bin/flink savepoint -d :savepointPath

This disposes the savepoint stored in `:savepointPath`.

-Note that since savepoints always go to a file system it is possible to also manually delete the savepoint via a regular file system operation. Keep in mind though that the savepoint only stores meta data that points to the actual checkpoint data. Therefore, if you manually want to delete a savepoint, you would have to include the checkpoint files as well. Since there is currently no straight forward way to figure out how a savepoint maps to a checkpoint, it is recommended to use the savepoint tool for this as described above.
+Note that since savepoints always go to a file system it is possible to also manually delete the savepoint via a regular file system operation. Up to Flink 1.2, this way a more tedious task which was solved with this tool.
— End diff –

Up to Flink 1.2, this was a more tedious task which was performed with the savepoint tool.

Github user EronWright commented on the issue:

https://github.com/apache/flink/pull/4024

Would you mind remarking as to whether it is safe to delete a savepoint after resuming a job from it?

Github user NicoK commented on the issue:

https://github.com/apache/flink/pull/4024

Deleting a savepoint will not affect other savepoints in case you are worried about this.
Other than that it depends on you and what you want to do with it, e.g. keeping it in case you want to resume from it at a later point in case of failures or so.

I did add a few notes which should make the first point clear.

Github user alpinegizmo commented on a diff in the pull request:

https://github.com/apache/flink/pull/4024#discussion_r119567341

— Diff: docs/setup/savepoints.md —
@@ -144,7 +144,7 @@ $ bin/flink savepoint -d :savepointPath

This disposes the savepoint stored in `:savepointPath`.

-Note that since savepoints always go to a file system it is possible to also manually delete the savepoint via a regular file system operation. Up to Flink 1.2, this way a more tedious task which was performed with the savepoint command above.
+Note that it is possible to also manually delete a savepoint via regular file system operations without affecting other savepoints or checkpoints (recall that each savepoint is self-contained). Up to Flink 1.2, this way a more tedious task which was performed with the savepoint command above.
— End diff –

... this was a more tedious ...

Github user alpinegizmo commented on the issue:

https://github.com/apache/flink/pull/4024

I read Eron's request a bit differently. Suppose you've taken a savepoint simply to manage a rescaling or redeployment. How quickly after resuming from the savepoint is it safe to delete it? Should one wait until the new job has been successfully checkpointed, for example?

Github user NicoK commented on the issue:

https://github.com/apache/flink/pull/4024

Personally, I would wait at least for another completed checkpoint before deleting a savepoint but from what I read, I would probably also always keep a savepoint around to recover from. I guess that depends on the availability needs and storage capacity that you have.
Maybe @StefanRRichter or @StephanEwen can elaborate a bit more on best practices there and extend the docs accordingly.

Github user zentol commented on the issue:

https://github.com/apache/flink/pull/4024

+1 to merge.

Github user zentol commented on the issue:

https://github.com/apache/flink/pull/4024

merging.

Github user asfgit closed the pull request at:

https://github.com/apache/flink/pull/4024

1.3: bfba2b1d453266f28523c47682d3cfe5ae67ce74
1.4: 60222a24123379bbbd044a6ca25817e64a9348b9