Issue 2205 - 4-3-stable #2208
Merged
Issue 2205 - 4-3-stable #2208
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
JustinKyleJames
force-pushed
the
issue_2205
branch
from
7218049 to
85f89ae
Compare
korydraughn
reviewed
Jul 10, 2024
s3_resource/include/irods/private/s3_resource/multipart_shared_data.hpp
Outdated
Show resolved
Hide resolved
s3_resource/include/irods/private/s3_resource/multipart_shared_data.hpp
Outdated
Show resolved
Hide resolved
s3_resource/include/irods/private/s3_resource/multipart_shared_data.hpp
Outdated
Show resolved
Hide resolved
s3_transport/include/irods/private/s3_transport/managed_shared_memory_object.hpp
Outdated
Show resolved
Hide resolved
JustinKyleJames
force-pushed
the
issue_2205
branch
from
f831fbc to
775e26f
Compare
korydraughn
reviewed
Jul 11, 2024
JustinKyleJames
force-pushed
the
issue_2205
branch
from
538f757 to
46b5905
Compare
Done |
|
One of your commits includes "#". Let's replace that with "number" so there's no confusion. |
JustinKyleJames
force-pushed
the
issue_2205
branch
from
46b5905 to
a03a62f
Compare
done |
|
If you're happy with it, pound it. |
korydraughn
approved these changes
Jul 11, 2024
JustinKyleJames
force-pushed
the
issue_2205
branch
from
a03a62f to
19a7c86
Compare
done |
|
Merging. |
|
@JustinKyleJames Now, we need to cherry-pick to the main branch, correct? |
Merged
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This pull request addresses two issues:
#2205 - For uploads the number of upload threads must be known. For PRC transfers with multiprocess uploads, only the first process is setting the keyword for the number of threads. Since these are multiple processes, this information needs to be saved in multiprocess shared memory so other processes know the correct number of transfer threads/processes.
I created a new shared data segment for the s3 resource which is separate from the s3 transport's shared memory. I renamed the transport's shared memory so that it is clear which is which.
To know when shared memory can be cleaned up, a counter was added to track the number of closes. Once the close counter reaches the number of threads the shared memory is cleaned up. I have verified that when running the test suite all shared memory files are cleaned up as expected.
Note that an up/down counter tracking create/open and closes will not work as the shared memory must be created before the first create/open and with an up/down counter this counter would immediately be zero and the shared memory would be deleted.
Read operations do not need this shared memory as we don't need to know the number of threads/processes in that case. In addition, the number of threads is not reliable in that scenario so it would be difficult to determine when the shared memory can be cleaned up.
I also learned via testing that our mechanism to calculate the necessary shared memory sizes is not accurate as the boost libraries need an unspecified amount of overhead memory. It was suggested that
100*sizeof(void*)bytes is plenty for this so I added this extra into both the new shared memory and the existing transport's shared memory.Finally, I wrapped most of the
get_number_of_threads_data_size_and_opr_type()processing within the shared memory'satomic_exec()as it simplifies everything if this part of the code is atomically executed. A lot of the changes below are simply a result of changing the indentation because of this.#2207 - I updated the code to use the test case name in filenames that are created.
Tests passed.