[jira] Updated: (TS-499) INKVIONTodoGet() and INKVIONBytesGet() returning obsense values ?

["Leif Hedstrom (JIRA)" <ji...@apache.org>]

     [ https://issues.apache.org/jira/browse/TS-499?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]

Leif Hedstrom updated TS-499:
-----------------------------

    Attachment: vio.diff

This patch fixes (I think) the 32-bit APIs to at least deal with the overflow and usage of INT64_MAX values to mean "all" or "unknown".

> INKVIONTodoGet() and INKVIONBytesGet() returning obsense values ?
> -----------------------------------------------------------------
>
>                 Key: TS-499
>                 URL: https://issues.apache.org/jira/browse/TS-499
>             Project: Traffic Server
>          Issue Type: Bug
>    Affects Versions: 2.0.0
>            Reporter: Leif Hedstrom
>            Priority: Critical
>             Fix For: 2.1.4
>
>         Attachments: vio.diff
>
>
> In a transform plugin, with chunked data in the origin response, we're seeing obscene return values from
> INKVIONTodoGet(input_vio);
> e.g. (with some new tracers) using v2.0.1:
> Oct 27 21:34:32.770] Server {139651219031824} DIAG: (null-transform) Entering handle_transform()
> [Oct 27 21:34:32.770] Server {139651219031824} DIAG: (null-transform) 	toWrite is 2147348222
> [Oct 27 21:34:32.770] Server {139651219031824} DIAG: (null-transform) 	avail is 2896
> This sort of works in the v2.0.1 branch, but fails completely on trunk (it'll typically completely hang the transaction, due to integer overflows). In my test, using trunk, "toWrite" becomes -1 for the chunked data.  A "working" solution is to not let the value overflow, but I think we should figure out why these numbers are so big in the first place (I'm guessing it's something related to 64-bit changes).
> There's a similar issue with INKVIONBytesGet() as well, and possible other APIs ?

-- 
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.