Weird HTTPS error

[Abilio Marques]

Hi,

I initially posted this error on the NGINX mailing list, but it looks to me
it has more to do with the developers.

The initial message is here:
http://mailman.nginx.org/pipermail/nginx/2018-March/055808.html

To make a long story short, I'm getting this message from NGINX 1.12.2:

[info] 1343#0: *55 SSL_read() failed (SSL:
error:1408F119:lib(20):func(143):reason(281)) while waiting for request,
client: 192.168.1.250, server: 0.0.0.0:443



Even when printed as info, this is very visible on the client side, as the
connection is severed before the data gets transferred.

There are 3 ways to make the problem "disappear", all discovered by
accident:
- Disable session resumption.
- Regenerate the key/certificate a number of times, until you get lucky.
- LD_PRELOAD even an "empty" (as in no C code) so library.

None of them are solutions, but just make the real problem hide.

This is happening on a MIPS. I'm just reposting here because I think
developers are the ones that might know where to start looking. I'm just
reposting here because I think developers are the ones that might know
where to start looking.

Any ideas on how to proceed?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.nginx.org/pipermail/nginx-devel/attachments/20180319/8f04d8dd/attachment.html>