alamb commented on issue #7251: URL: https://github.com/apache/arrow-rs/issues/7251#issuecomment-2738313884
> > [@crepererum](https://github.com/crepererum) rightly pointed out that implementing retries (aka [#7242](https://github.com/apache/arrow-rs/issues/7242)) would be better than splitting into smaller requests to make a timeout as the retry mechanism automatically adjusts to current network conditions > > Isn't there an upper bound on the timeout (30s by default)? And if the bound isn't large enough to push that 200MiB row group through a slow connection, won't the request fail anyway? I think the idea is you don't re-request the entire object, only bytes remaining So let's say you had a 200 MB request but the network can only retrieve 10MB in 30s - The first request would fetch the first 10MB but timeout - Then the retry would request the remaining 190MB - The second request would fetch the second 10MB and timeout - Then the retry would request the remaining 180MB - .. and so on I agree this is not clear -- I will post the same on https://github.com/apache/arrow-rs/issues/7242 -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: [email protected] For queries about this service, please contact Infrastructure at: [email protected]
