Andrew Clayton

Sure, can do. There is also *header_buffer_size* that is _not_ currently settable by the user, should it be?

Hmm, there is also ``` proxy_header_buffer_size proxy_buffer_size proxy_buffers proxy_timeout proxy_send_timeout proxy_read_timeout ``` *proxy_buffers* seems to be unused. These are all set in nxt_router.c::nxt_router_conf_create() along with the other http values. However...

Dropped the 'Router: remove unused proxy_buffers structure member.' commit as that is perhaps better done as a separate pull-request.

@tagur87 Can you test the below patch? ```diff diff --git a/src/nxt_conf_validation.c b/src/nxt_conf_validation.c index fe6c22e5..fde0644f 100644 --- a/src/nxt_conf_validation.c +++ b/src/nxt_conf_validation.c @@ -312,6 +312,9 @@ static nxt_conf_vldt_object_t nxt_conf_vldt_http_members[] = { }, {...

@tagur87 Thanks for testing! I will see about getting this patch through the review process and I will put this patch in a new pull request where it can be...

Closing this as the source branch (again) only contains the single 'large_header_buffer_size' patch...

[PR](https://github.com/nginx/unit/pull/771) for the afore mentioed patch.

I also see the same thing by simply doing (on Fedora 35, jemalloc 5.2.1, php 8.0.18) ``` $ LD_PRELOAD=/lib64/libjemalloc.so.2 php free(): invalid pointer Aborted (core dumped) ```

capabilities(7) might be a nice way of solving the unlink(2)ing of the socket files problem, on Linux at least. I don't know if they ever made it to other Unix...

Heh, I think that's a little extreme. It allows a process to override file system permissions yes, but it's not as bad as CAP_SYS_ADMIN (that's the basically root one) and...