Implement filtering mechanism for cmetrics context #193
Conversation
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
cosmo0920
force-pushed
the
cosmo0920-cmt_filter
branch
from
fa21f6e
to
a140738
Compare
cosmo0920
changed the title
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
cosmo0920
force-pushed
the
cosmo0920-cmt_filter
branch
from
3071c75
to
05baacd
Compare
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Because (con)cat(enate) implies copying the pointer/context. Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
|
I've done for the suggested changes. |
src/cmt_filter.c
Outdated
| struct cmt_map_label *label; | ||
| size_t label_key_size = 0; | ||
|
|
||
| if (label_key != NULL) { |
There was a problem hiding this comment.
if there is a chance this can be NULL the code below will not work. Either the check is not necessary or we just return with -1 if is NULL
There was a problem hiding this comment.
There is a chance to become NULL. Because we are able to specify an external function only here.
And this should be injected via compare_ctx and callback function.
There was a problem hiding this comment.
label_key is only needed for prefix matching case. I added a clause which checks for CMT_FILTER_PREFIX flag.
src/cmt_filter.c
Outdated
| * The length of string is changed by the callback and not determined by fqname. | ||
| */ | ||
| if (compare_ctx != NULL && compare != NULL) { | ||
| return compare(compare_ctx, src->fqname, strlen(src->fqname)); |
This comment was marked as outdated.
This comment was marked as outdated.
Sorry, something went wrong.
This comment was marked as resolved.
This comment was marked as resolved.
Sorry, something went wrong.
There was a problem hiding this comment.
Once I rethought this, I realized this could be calculated before it's needed. So, prefix matching case and using an external function case should be able to calculate size of fqname. So, the previous my comment is wrong. Thanks for your suggestion.
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
If label_key is NULL and CMT_FILTER_PREFIX specified case, SEGV will happen. So, we need to return imetteately for this case. Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
cosmo0920
force-pushed
the
cosmo0920-cmt_filter
branch
from
3a3dcc6
to
58f8951
Compare
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
Signed-off-by: Hiroshi Hatake <hiroshi@chronosphere.io>
I implemented filtering mechanism which is mainly targeted for full qualified name of metrics (fqname).
Maybe, we need to implement filtering mechanism for label values but traversing values of labels are more complex rather than just searching with keys of labels.
Also, cmetrics should provide a capability to inject external callback function which uses Onigmo contexts that are mainly used in filter_grep or perhaps processor_grep?