[
https://issues.apache.org/jira/browse/PIG-1001?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=12771072#action_12771072
]
Alan Gates commented on PIG-1001:
---------------------------------
I have a question on this code in JobControlCompiler.updateMROpPlan:
{code}
for (Job job : completeFailedJobs) // remove all subsequent jobs
{
List<MapReduceOper> fifo = new ArrayList<MapReduceOper>();
fifo.add(jobMroMap.get(job));
while (!fifo.isEmpty())
{
MapReduceOper mro = fifo.remove(0);
List<MapReduceOper> succs = plan.getSuccessors(mro);
if (succs != null)
fifo.addAll(succs);
plan.remove(mro);
numRemoved++;
}
}
{code}
If we have a dependency graph like: A->B->C and A fails, won't the above code
only remove B and not C? OperatorPlan.getSuccessors() only gets immediate
successors, not all successors. I think you want OperatorPlan.trimBelow()
instead.
> Generate more meaningful error message when one input file does not exist
> -------------------------------------------------------------------------
>
> Key: PIG-1001
> URL: https://issues.apache.org/jira/browse/PIG-1001
> Project: Pig
> Issue Type: Bug
> Affects Versions: 0.4.0
> Reporter: Daniel Dai
> Fix For: 0.6.0
>
> Attachments: PIG-1001-1.patch
>
>
> In the following query, if 1.txt does not exist,
> a = load '1.txt';
> b = group a by $0;
> c = group b all;
> dump c;
> Pig throws error message "ERROR 2100: file:/tmp/temp155054664/tmp1144108421
> does not exist.", Pig should deal with it with the error message "Input file
> 1.txt not exist" instead of those confusing messages.
--
This message is automatically generated by JIRA.
-
You can reply to this email to add a comment to the issue online.
