diff options
author | Tor Brede Vekterli <vekterli@verizonmedia.com> | 2021-01-06 15:12:33 +0000 |
---|---|---|
committer | Tor Brede Vekterli <vekterli@verizonmedia.com> | 2021-01-06 15:12:33 +0000 |
commit | eaa27facad21cab7d37adbc6e041eb0680b49bca (patch) | |
tree | 79bab0549c8398acbb6f9c4ed7e748da1a5d5c17 /storage | |
parent | 55042393392a94095d5c821e8118d06a7067e048 (diff) |
Increase number of bucket info failures before a warning is printed
Was >=2 seconds, now >=20 seconds. 2 seconds may cause spurious log
warnings on heavily loaded nodes that lag behind the distributor for
some seconds before being able to switch over to a new cluster state
version.
Diffstat (limited to 'storage')
-rw-r--r-- | storage/src/vespa/storage/distributor/pendingclusterstate.h | 2 |
1 files changed, 1 insertions, 1 deletions
diff --git a/storage/src/vespa/storage/distributor/pendingclusterstate.h b/storage/src/vespa/storage/distributor/pendingclusterstate.h index f79a3185c67..c355af3e00f 100644 --- a/storage/src/vespa/storage/distributor/pendingclusterstate.h +++ b/storage/src/vespa/storage/distributor/pendingclusterstate.h @@ -157,7 +157,7 @@ public: private: // With 100ms resend timeout, this requires a particular node to have failed // for _at least_ threshold/10 seconds before a log warning is emitted. - constexpr static size_t RequestFailureWarningEdgeTriggerThreshold = 20; + constexpr static size_t RequestFailureWarningEdgeTriggerThreshold = 200; /** * Creates a pending cluster state that represents |