Re: [pbs-devel] [PATCH proxmox-backup] sync job: pull: avoid blocking future during thread pool completion

[Christian Ebner <c.ebner@proxmox.com>]

On 10/14/25 1:25 PM, Fabian Grünbichler wrote:
> On October 9, 2025 11:20 am, Christian Ebner wrote:
>> Chunk insertion during a pull sync job is performed via a thread pool
>> to increase performance. The caller waits for all threads to be
>> completed at the end, relying on the respective associated threads to
>> return from `std::thread::JoinHandle::join`.
>>
>> Insertion into the chunk store can however take time, especially
>> for chunks being uploaded to the object store for datastores with
>> S3 backend.
>>
>> To avoid blocking the async runtime, spawn a dedicated task to await
>> the thread pool completion.
>>
>> Signed-off-by: Christian Ebner <c.ebner@proxmox.com>
>> ---
>> Encountered while looking for possible causes of proxy hangs as reported
>> in https://forum.proxmox.com/threads/171422/post-807098
>>
>>   src/server/pull.rs | 2 +-
>>   1 file changed, 1 insertion(+), 1 deletion(-)
>>
>> diff --git a/src/server/pull.rs b/src/server/pull.rs
>> index 08e6e6b64..53f1a570d 100644
>> --- a/src/server/pull.rs
>> +++ b/src/server/pull.rs
>> @@ -232,7 +232,7 @@ async fn pull_index_chunks<I: IndexFile>(
>>   
>>       drop(verify_and_write_channel);
>>   
>> -    verify_pool.complete()?;
>> +    let _ = tokio::task::spawn_blocking(|| verify_pool.complete()).await?;
> 
> doesn't this change error handling semantics? the inner Result is no
> longer handled, AFAICT..

Oh, you are right! This now only covers the JoinError... Will send a v2 
fixing this shortly.


_______________________________________________
pbs-devel mailing list
pbs-devel@lists.proxmox.com
https://lists.proxmox.com/cgi-bin/mailman/listinfo/pbs-devel