[mod_gnutls-devel] Apache 2 taking 100% CPU due to mod-gnutls

Sunil Mohan Adapa sunil at medhas.org
Wed May 25 20:54:31 CEST 2016


On 05/25/2016 10:44 PM, Thomas Klute wrote:
[...]
> Thank you for the report! Unfortunately I could not reproduce the bug,
> but I agree that mgs_transport_write not setting errno might be an
> issue. Could you check if the attached patch fixes the problem?

The patch fixes the problem.  Tested on Debian package 0.7.4-2:

- Without the patch, the problem is reproducible consistently with 'ab
-c 8 -n 800' even before it finishes 100 requests. This happened every
time I ran (I ran at least 5 times).

- I have been unable to reproduce the problem with 'ab -c 8 -n 8000' and
with 5 more shorter runs.

> 
> If this does not help, would you be able to provide me with a minimal
> configuration that can reproduce the problem (e.g. Apache conf, CGI
> script for the slow proxy backend, shell script to send requests)? Just
> the information which client you are using might be helpful as well
> because your report looks like a bug that might be exposed by specific
> client behavior at end of connection.

I can still work on minimal setup if necessary.  Let me know.

Thank you,

-- 
Sunil

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: </pipermail/attachments/20160526/3822ac6e/attachment.sig>


More information about the mod_gnutls-devel mailing list