Andrii Chubatiuk
Andrii Chubatiuk
> i don't grok the "batch" knobs. Why do we need those? they are not useful in case of small logs stream, but for huge one these settings should allow...
> batches of what? batches of journal entries > why do we have to care? server can reach connections limit too fast if lot's of servers start uploading uncontrollable batches...
@yuwata divided into multiple commits
hey @yuwata moved compression related changes from [this](https://github.com/systemd/systemd/pull/34595) PR here
@yuwata please let's think about it in a separate PR as: - adding Accept-Encoding support doesn't allow to have a proper CompressionLevel setting as compression level boundaries are different for...
> I think journal-upload should not send any compressed data unless the client side send Accept-Encoding tag. first batch can contain a biggest amount of data, for which compression could...
@yuwata added content encoding negotiation, but I'm parsing just a first Accept-Encoding header and do not take into in account `q=` values as journal-remote decrements `q=` value for each subsequent...
@yuwata added content encoding negotiation, but disabled it for centos 9, which has too old libcurl, which complicates response headers parsing
> needs test case @poettering what kind of test case?
@yuwata @poettering could you please take a look?