Updated testing and fixed bug in SSLSNIConfig #12248
Conversation
|
Looking at the updated docs globs in the middle are no longer supported. It adds the entry with the asterisk in the middle to the exact match list with no warning. |
| if (this != &other) { | ||
| match = std::move(other.match); | ||
| inbound_port_ranges = std::move(other.inbound_port_ranges); | ||
| rank = other.rank; |
There was a problem hiding this comment.
This is the bug fix. ⬆️
The rest of the changes are more tests and an updated config.
|
Hey @masaori335, you added in the performance enhancements and ranking to begin with, so you should know more about this code than anyone else. Can you please review the bug fix and the added tests? If you can think of more tests I can add them too. I have another PR #12219 to change the PCRE regexes to use the I have to check to see if we are covering case were we verify the capture group. |
|
@masaori335 Can you take a look at this? I would like to get it merged in, I have another PR that relies on this one. |
| http2: false | ||
|
|
||
| # test glob in the middle, this will be an exact match | ||
| - fqdn: "cat.*.com" |
There was a problem hiding this comment.
Discussion around the #9736, we concluded don't allow this. This should be error.
- Allow single left-most
*in the fqdn field
https://docs.trafficserver.apache.org/en/latest/admin-guide/files/sni.yaml.en.html
There was a problem hiding this comment.
This is for these tests for invalid?
https://github.com/apache/trafficserver/pull/12248/files#diff-40f6583cfe88d8e4ea494c95bf3dbceed8a172c40836afe607f9ff7c1981e3a8R160-R171
|
The main change seems reasonable, but the wildcard in middle is we don't support. If you rush, we can merge this and deal with the issue later. |
|
Cherry-picked to 10.1.x branch |
(cherry picked from commit 4d12805)
3 participants
If the
sni.yamlconfiguration file contained multiple wildcard entries, an incorrect wildcard entry could be matched instead of the exact match listed first in the file.This issue occurred because the
rankfor wildcard entries was set to 0 (the default value) and was not updated in the move constructor when the list was resized to accommodate a new entry.PR that caused the bug: #9736
TODO: