Michael Ho has posted comments on this change. ( 
http://gerrit.cloudera.org:8080/8100 )

Change subject: IMPALA-5940: Avoid log spew by using Status::Expected.
......................................................................


Patch Set 3:

(2 comments)

http://gerrit.cloudera.org:8080/#/c/8100/3/be/src/runtime/client-cache.h
File be/src/runtime/client-cache.h:

http://gerrit.cloudera.org:8080/#/c/8100/3/be/src/runtime/client-cache.h@246
PS3, Line 246:         return 
Status::Expected(ErrorMsg(TErrorCode::RPC_RECV_TIMEOUT, err));
> Yea, the backtrace is not all that useful here. Forgot to add that. I was a
I agree that removing the backtrace here is the right direction as 90% of the 
cases logged are from DataStreamSender and they are considered "harmless" in 
this case.

The caller of this function from DataStreamSender is from  
DataStreamSender::Channel::DoTransmitDataRpc() which will react to a timeout 
error by retrying the recv part of the RPC repeatedly until success or a 
non-timeout related error is found. So, in that sense, the RECV_TIMEOUT error 
shouldn't be logged in GetSendStatus() until it hits other errors.

I did a quick audit of all callers of DoRpc() and it appears that some callers 
didn't really log the error. So, it may still be a bit inconvenient in those 
cases but this can be made up by adding more logging in those cases instead. 
(e.g. CatalogOpExecutor::Exec)


http://gerrit.cloudera.org:8080/#/c/8100/3/be/src/runtime/client-cache.h@281
PS3, Line 281:         return 
Status::Expected(ErrorMsg(TErrorCode::RPC_RECV_TIMEOUT, err));
> same set of comment/question
This one is definitely fine to not log the backtrace as we know this is coming 
from DataStreamSender for sure.



--
To view, visit http://gerrit.cloudera.org:8080/8100
To unsubscribe, visit http://gerrit.cloudera.org:8080/settings

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: I38088482377a1c3e794a9c8178ef83f29957a330
Gerrit-Change-Number: 8100
Gerrit-PatchSet: 3
Gerrit-Owner: Philip Zeyliger <[email protected]>
Gerrit-Reviewer: Alex Behm <[email protected]>
Gerrit-Reviewer: Dan Hecht <[email protected]>
Gerrit-Reviewer: Michael Ho <[email protected]>
Gerrit-Reviewer: Mostafa Mokhtar <[email protected]>
Gerrit-Reviewer: Philip Zeyliger <[email protected]>
Gerrit-Reviewer: Sailesh Mukil <[email protected]>
Gerrit-Comment-Date: Tue, 03 Oct 2017 18:29:16 +0000
Gerrit-HasComments: Yes

Reply via email to