Github user tillrohrmann commented on the pull request:
https://github.com/apache/flink/pull/1227#issuecomment-145810928
Good PR @uce. I think the proposal in your description to let the
client-side (TMs) 1. check whether they have the file cached, 2. check the
filesystem backend and 3. (this case should then never happen in recovery mode)
ask the JM, in order to obtain the required Jars would be a good improvement.
Currently, I've got the feeling that the file state backend is too tightly
coupled with the `BlobServer`. IMHO, it would be better to add an abstraction
so that the effective backend to distribute the Jars can be easily swapped.
Furthermore, I couldn't find a check whether the user provided
`STATE_BACKEND_FS_RECOVERY_PATH` path points actually to a distributed file
system and is, thus, accessible by the TMs. Maybe we could add a check which,
if false, will give a comprehensive warning.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at infrastruct...@apache.org or file a JIRA ticket
with INFRA.
---
