[jira] [Commented] (HAWQ-1282) Shared Input Scan may result in endless loop.

["Hubert Zhang (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/HAWQ-1282?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15834026#comment-15834026 ] 

Hubert Zhang commented on HAWQ-1282:
------------------------------------

Here are some context.
After running some queries(including shared input scan) There may be residual process left. The call stack of residual process is as follows:
```
#0  0x00000032214e1523 in select () from /lib64/libc.so.6
#1  0x000000000069a072 in writer_wait_for_acks (share_id=0, xslice=Unhandled dwarf expression opcode 0xf3
) at nodeShareInputScan.c:885
#2  0x000000000069ae92 in shareinput_writer_notifyready (share_id=0, xslice=3, planGen=PLANGEN_PLANNER)
    at nodeShareInputScan.c:854
#3  0x0000000000693f48 in ExecMaterial (node=0x2e93a28) at nodeMaterial.c:248
#4  0x000000000066df31 in ExecProcNode (node=0x2e93a28) at execProcnode.c:971
#5  0x0000000000699c17 in init_tuplestore_state (node=0x2e935c8) at nodeShareInputScan.c:100
#6  0x0000000000699f99 in ShareInputNext (node=0x2e935c8) at nodeShareInputScan.c:210
#7  0x000000000066dd51 in ExecProcNode (node=0x2e935c8) at execProcnode.c:1003
#8  0x000000000069bd50 in ExecSort (node=0x2e93020) at nodeSort.c:335
```
Function writer_wait_for_acks() will be in an endless loop if every time the select function returns 0 because of  timeout or minus value.

> Shared Input Scan may result in endless loop.
> ---------------------------------------------
>
>                 Key: HAWQ-1282
>                 URL: https://issues.apache.org/jira/browse/HAWQ-1282
>             Project: Apache HAWQ
>          Issue Type: Bug
>          Components: Core
>            Reporter: Hubert Zhang
>            Assignee: Hubert Zhang
>             Fix For: 2.1.0.0-incubating
>
>
> There are residual process after running some queries.  Through the call stack, we find that there is an endless loop in function writer_wait_for_acks() in shared input scan.
> We plan to add max retry times to avoid this problem.
> Also, there are some file handler leaks in retry_read() and retry_write() of shared input scan.



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