rgw: civetweb fixes for v1.1 upgrade #21123
Conversation
src/rgw/rgw_civetweb_frontend.cc
Outdated
| @@ -55,7 +55,7 @@ int RGWCivetWebFrontend::run() | |||
| std::to_string(g_conf->rgw_thread_pool_size)); | |||
| set_conf_default(conf_map, "decode_url", "no"); | |||
| set_conf_default(conf_map, "enable_keep_alive", "yes"); | |||
| set_conf_default(conf_map, "validate_http_method", "no"); | |||
| //set_conf_default(conf_map, "validate_http_method", "no"); | |||
There was a problem hiding this comment.
TODO: drop this once we decide whether it is necessary
theanalyst
force-pushed
the
wip-civetweb-1.1
branch
from
88a222d
to
0d56c8a
Compare
|
Initial qe run: http://pulpito.ceph.com/abhi-2018-04-04_15:05:26-rgw-wip-civetweb-1.1-distro-basic-smithi/ We have failures in s3-tests with rfc2616 and swift failures (due to pip?), dropping the rfc2616 checks and rerunning as http://pulpito.ceph.com/abhi-2018-04-05_17:09:22-rgw-wip-civetweb-1.1-distro-basic-smithi/ The above suite failed for a few swift suites, re introduced the earlier downstream method validation drop and also added a downstream commit to allow unicode urls (civetweb was just allowing url encoded urls, however our swift tests sends both url encoded and non url encoded unicode urls), the new suite is running at http://pulpito.ceph.com/abhi-2018-04-09_17:18:28-rgw-wip-civetweb-1.1-distro-basic-smithi/ rerun of the failed jobs at http://pulpito.ceph.com/abhi-2018-04-11_08:17:08-rgw-wip-civetweb-1.1-distro-basic-smithi/, unfortunately the jobs failed due to env failures again, the valgrind failures seem to be the multisite ones, though the failures itself are with beast as a frontend and not running civetweb |
|
I have updated the ceph/civetweb#24 to correct branches. This PR needs to reference the correct sha1 once that merges |
theanalyst
force-pushed
the
wip-civetweb-1.1
branch
from
204bcfb
to
ec2dbca
Compare
|
jenkins test this please |
|
jenkins test this please (submodule sha not found as the remote didn't exist in ceph/civetewb) |
|
@cbodley @mattbenjamin I have pushed the last changeset based on version CMake tells us, should we do it this way or go the objdump route? |
i think the cmake route is fine for mimic. let's revisit the linking strategy with @mdw-at-linuxbox later |
theanalyst
changed the title
|
err that didn't work, version_greater_equal was introduced in cmake 3.7 only, I'll rewrite to do a greater than or equal to myself |
Introduces the following additions in rgw: - allow_unicode_in_urls introduced with a corresponding downstream commit in civetweb, as the newer version of civetweb validates that urls are url encoded which swifttests do not follow, so introduce this as a configurable which we set as true - mg header struct changes in civetweb update, use auto here - drop info->uri and use local_uri instead as the former is deprecated wip: rgw: civetweb fixes for v1.1 upgrade Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
Since newer versions of civetweb are also strict on rfc2616 checks let's enforce strict rfc2616 checks in s3tests Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
While not ideal, we set the civetweb openssl 1.1 conditional compile based on the openssl version that cmake reports. In future we should make civetweb itself do this based on OPENSSL_VERSION_COMPAT Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
Updating the sha1 to track ceph-master which currently updates civetweb to 1.1 Signed-off-by: Abhishek Lekshmanan <abhishek@suse.com>
theanalyst
force-pushed
the
wip-civetweb-1.1
branch
from
b999bf3
to
823ce04
Compare
|
The submodule change probably requires @mattbenjamin / @yehudasa approval |
(update: this was readded as swift tests depend on this)
This changeset does the following:
validate_http_method, needs discussion