Tim Armstrong has posted comments on this change. Change subject: IMPALA-5715: (mitigation only) defer destruction of MemTrackers ......................................................................
Patch Set 4: (3 comments) http://gerrit.cloudera.org:8080/#/c/7492/3/be/src/runtime/mem-tracker.cc File be/src/runtime/mem-tracker.cc: PS3, Line 114: if (closed_) return; > Why this change ? Why not always have the DCHECK() below ? I don't think the DCHECK necessarily makes sense if there's a consumption metric. E.g. if it's the TCMalloc consumption, it's probably not going to be exactly zero. I guess we never Close() a MemTracker with a consumption metric so it's academic at the moment. I can remove the condition if you think it's clearer. PS3, Line 200: mTracker::~MemTracker() { > Ideally, we'd like to have a DCHECK(closed_) here if we fix up all backend Yeah I agree. I went through and looked at the tests and it seemed like a lot of added code for not much benefit. I made this a little stricter by checking whether it's running in a backend test. http://gerrit.cloudera.org:8080/#/c/7492/3/be/src/runtime/query-state.cc File be/src/runtime/query-state.cc: Line 80: // and they can be safely destroyed. > nit: long line Done -- To view, visit http://gerrit.cloudera.org:8080/7492 To unsubscribe, visit http://gerrit.cloudera.org:8080/settings Gerrit-MessageType: comment Gerrit-Change-Id: I205abb0076d1ffd08cb93c0f1671c8b81e7fba0f Gerrit-PatchSet: 4 Gerrit-Project: Impala-ASF Gerrit-Branch: master Gerrit-Owner: Tim Armstrong <[email protected]> Gerrit-Reviewer: Henry Robinson <[email protected]> Gerrit-Reviewer: Matthew Jacobs <[email protected]> Gerrit-Reviewer: Michael Ho <[email protected]> Gerrit-Reviewer: Tim Armstrong <[email protected]> Gerrit-HasComments: Yes
