Dan Hecht has posted comments on this change.

Change subject: IMPALA-2399: Check for mem limit in allocations in parquet 
scanner and decompressor.
......................................................................


Patch Set 7:

(6 comments)

http://gerrit.cloudera.org:8080/#/c/2203/7/be/src/common/status.cc
File be/src/common/status.cc:

Line 20: #include "runtime/runtime-state.h"
it seems a bit unfortunate to add the dependency in this direction since status 
is such a low-level primitive. What do you think about instead making the new 
method a member of MemTracker or RuntimeState?


http://gerrit.cloudera.org:8080/#/c/2203/7/be/src/exec/hdfs-parquet-scanner.cc
File be/src/exec/hdfs-parquet-scanner.cc:

Line 755:           len, "StringValue");
nit: get rid of 'details' variables, but okay to ignore if you prefer it this 
way.


Line 757:           parent_->state_, details, (int64_t)len);
i agree its better to not introduce more SetMemLimitExceeded() calls since we 
want to get rid of that as part of this work.


http://gerrit.cloudera.org:8080/#/c/2203/7/be/src/runtime/runtime-state.cc
File be/src/runtime/runtime-state.cc:

Line 244: void RuntimeState::LogFailedAllocDetails(const MemTracker* tracker,
"Log" could mean LOG/VLOG rather than LogError() and failed allocations are not 
exactly the same as mem limit exceeded (they could happen for other reasons), 
so how about calling this thing: LogErrorMemLimitExceeded()?


Line 268:   }
not for this change, but I wonder if the stuff we pass to LogError() should 
instead become detail to the MemLimitExceeded status itself.


Line 273:   DCHECK_GE(failed_allocation_size, 0);
this DCHECK could be in the Log function.


-- 
To view, visit http://gerrit.cloudera.org:8080/2203
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-MessageType: comment
Gerrit-Change-Id: Ic70400407b7662999332448f4d1bce2cc344ca89
Gerrit-PatchSet: 7
Gerrit-Project: Impala
Gerrit-Branch: cdh5-trunk
Gerrit-Owner: Michael Ho <[email protected]>
Gerrit-Reviewer: Dan Hecht <[email protected]>
Gerrit-Reviewer: Michael Ho <[email protected]>
Gerrit-Reviewer: Skye Wanderman-Milne <[email protected]>
Gerrit-Reviewer: Tim Armstrong <[email protected]>
Gerrit-HasComments: Yes

Reply via email to