Christopher Faulet

Results 470 comments of Christopher Faulet

Indeed, since the 2.6, unknown encoding (everything except "chunked") are rejected and multiple "chunked" values too. Empty values are also rejected. So concretely, it is not possible to have more...

@jvinolas, I'm closing the issue but if you have new info, feel free to reopen it. Thanks !

I quickly reviewed your patch and it seems good to me (but I have not tested it). Could you make a clean patch following the CONTRIBUTING rules ?

I'm able to reproduce the bug. It seems to be a problem with private backend connections. It is probably because of the commit 04a24c5ea. The function `srv_del_conn_from_list()` may be called...

But the Willy's comment remains true. All this stuff is a bit cryptic and relying on connection flags to known if a connection is in a list or not (include...

@wtarreau, shall we close the issue or not ? (I have not read all the history, sorry ...)

I guess you can set the severity output format instead. It can be set dynamically on the CLI with `set severity-output` command or statically in the config file via `severity-output`...

Well, it is not forcefully always obvious. But most of reasons for a `PR` termination state can be determined with the status code: * `PR/500`: it is a rewrite error...

> If I understand it correctly: > > 1. exceeding the configured header limit would be counted as a request parsing error, and > 2. not having a status code...