impala-reviews mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Philip Zeyliger (Code Review)" <>
Subject [Impala-ASF-CR] IMPALA-6285: Don't print stack trace on RPC errors.
Date Thu, 07 Dec 2017 04:19:41 GMT
Philip Zeyliger has posted comments on this change. (

Change subject: IMPALA-6285: Don't print stack trace on RPC errors.

Patch Set 1:


Seems reasonable to me. Have you checked that timeouts actually show the RPC name and client

The only test that I saw that exercised DoRpc was be/src/runtime/; would
it make sense to assert something about the status on errors?
File be/src/exec/
PS1, Line 279:       request, &response, "GetCatalogObject"));
We probably frown upon it, but it may be that typeId(request) inside of DoRpc() happens at
compile time and avoids the need for changing all of these. On balance, I think explicit (what
you have) is better, but I figured I'd ask about our style preferences there.
File be/src/runtime/client-cache.h:
PS1, Line 298:   /// Function for returning an error Status for general RPC errors not due
to recv
nit: "Function for" can be removed (and tenses changed slightly); also in line 310.

To view, visit
To unsubscribe, visit

Gerrit-Project: Impala-ASF
Gerrit-Branch: master
Gerrit-MessageType: comment
Gerrit-Change-Id: Ia83294494442ef21f7934f92ba9112e80d81fa58
Gerrit-Change-Number: 8788
Gerrit-PatchSet: 1
Gerrit-Owner: Michael Ho <>
Gerrit-Reviewer: Mostafa Mokhtar <>
Gerrit-Reviewer: Philip Zeyliger <>
Gerrit-Reviewer: Sailesh Mukil <>
Gerrit-Comment-Date: Thu, 07 Dec 2017 04:19:41 +0000
Gerrit-HasComments: Yes

  • Unnamed multipart/alternative (inline, 8-Bit, 0 bytes)
View raw message