[kudu] branch master updated: external_mini_cluster: bump start process timeout a bit more

[ad...@apache.org]

This is an automated email from the ASF dual-hosted git repository.

adar pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git


The following commit(s) were added to refs/heads/master by this push:
     new 9664526  external_mini_cluster: bump start process timeout a bit more
9664526 is described below

commit 966452659e40b5cb699860b4124dc7714ee895a0
Author: Adar Dembo <ad...@cloudera.com>
AuthorDate: Thu May 30 23:13:29 2019 -0700

    external_mini_cluster: bump start process timeout a bit more
    
    Commit 08b916d4b did help reduce the occurrence of NTP desynchronization in
    tests, but not completely. Worse, tests that now do time out do so in a way that
    doesn't blame NTP at all. For example:
    
      Bad status: Timed out: failed to start masters: Unable to start Master
      at index 0: Timed out after 60.000s waiting for process (...) to write
      info file (...)
    
    This also affects test result reporting, which tries to avoid reporting test
    runs where NTP failed to synchronize by searching for the "Clock considered
    unsynchronized" magic string. Of course, that won't happen if the minicluster,
    thinking that its subprocesses timed out, kills them before they fail that way.
    
    Here's a dumb fix: let's increase the start process timeout a bit. That gives
    the servers more time with which to reach the NTP desynchronization timeout.
    
    Change-Id: I95de5368298b9c7fd1ea0b7166dd6c3cd24e356b
    Reviewed-on: http://gerrit.cloudera.org:8080/13484
    Tested-by: Kudu Jenkins
    Reviewed-by: Will Berkeley <wd...@gmail.com>
---
 src/kudu/mini-cluster/external_mini_cluster.cc | 2 +-
 src/kudu/mini-cluster/external_mini_cluster.h  | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/kudu/mini-cluster/external_mini_cluster.cc b/src/kudu/mini-cluster/external_mini_cluster.cc
index 73145ba..b60fcb2 100644
--- a/src/kudu/mini-cluster/external_mini_cluster.cc
+++ b/src/kudu/mini-cluster/external_mini_cluster.cc
@@ -111,7 +111,7 @@ ExternalMiniClusterOptions::ExternalMiniClusterOptions()
       hms_mode(HmsMode::NONE),
       enable_sentry(false),
       logtostderr(true),
-      start_process_timeout(MonoDelta::FromSeconds(60)),
+      start_process_timeout(MonoDelta::FromSeconds(70)),
       rpc_negotiation_timeout(MonoDelta::FromSeconds(3)) {
 }
 
diff --git a/src/kudu/mini-cluster/external_mini_cluster.h b/src/kudu/mini-cluster/external_mini_cluster.h
index e038ec7..f7f4caa 100644
--- a/src/kudu/mini-cluster/external_mini_cluster.h
+++ b/src/kudu/mini-cluster/external_mini_cluster.h
@@ -167,7 +167,7 @@ struct ExternalMiniClusterOptions {
   // Amount of time that may elapse between the creation of a daemon process
   // and the process writing out its info file.
   //
-  // Default: 60s.
+  // Default: 70s (just a bit more than --ntp_initial_wait_secs).
   MonoDelta start_process_timeout;
 
   // Parameter for the cluster's RPC messenger: timeout interval after which