[jira] [Commented] (HTRACE-328) htraced continues scanning in some cases even when no more results are possible

["Masatake Iwasaki (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HTRACE-328?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15057847#comment-15057847 ] 

Masatake Iwasaki commented on HTRACE-328:
-----------------------------------------

I was puzzled at first look but the cause seems to be that {{break}} can not be used under {{case}} to break out of surrounding {{for}} loop.. The patch looks good overall but one nit.

{code}
        case common.GREATER_THAN_OR_EQUALS:
                if bytes.Compare(val, pred.key) >= 0 {
                        return SATISFIED
                } else {
-                       return NOT_SATISFIED
+                       return NOT_YET_SATISFIED
                }
{code}

Is this change needed? Tests passed without this.


> htraced continues scanning in some cases even when no more results are possible
> -------------------------------------------------------------------------------
>
>                 Key: HTRACE-328
>                 URL: https://issues.apache.org/jira/browse/HTRACE-328
>             Project: HTrace
>          Issue Type: Bug
>          Components: htraced
>    Affects Versions: 4.1
>            Reporter: Sean Busbey
>            Assignee: Colin Patrick McCabe
>            Priority: Critical
>         Attachments: HTRACE-328.001.patch
>
>
> htraced continues scanning in some cases even when no more results are possible.
> The clearest example of this is doing a point query on a span id.  It should be just seeking to where the span would be (if it existed) and then either reading it or returning "not found."  Instead, it scans the whole leveldb database.  This can be quite slow on larger leveldb instances.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)