tree c2e84a33752770342c5efca3f7bcaeb15171391d
parent 25e06e17fae9fc614f22b045f967f0e7492b7dd4
author Olav Loite <koloite@gmail.com> 1566400116 +0200
committer Knut Olav Løite <koloite@gmail.com> 1566895992 +0000

Spanner: prevent rpc timeouts in replenishPool

The RPC's in replenishPool used the context with timeout that was
passed in by the maintainer method. By default, this would have a
timeout of 1 minute, but can be configured to a different value.
This timeout value should not be used to timeout the gRPC calls,
as a very low value would start timing out all or most gRPC calls
for creating and deleting sessions in replenishPool, causing the
method to report a lot of timeout error and preventing it from
executing its actual task. The TestStressSessionPool tests the
session pool with different configurations. One of them is with a
healthCheckSampleInterval of 10ms, which currently fills up the
test log with a large number of timeout errors (when executed
with the -race argument).

Change-Id: I47bb58185f2f9768dc96aa50e2c420cbe31bac7a
Reviewed-on: https://code-review.googlesource.com/c/gocloud/+/44177
Reviewed-by: kokoro <noreply+kokoro@google.com>
Reviewed-by: Jean de Klerk <deklerk@google.com>