From 9bc6055af8e881635ceec9cbe5df51677cf945bc Mon Sep 17 00:00:00 2001 From: Ronnie Sahlberg Date: Tue, 20 Mar 2012 11:38:20 +1100 Subject: DOC document the FetchCollapse tunable (This used to be ctdb commit c37aa6f3738693653f64c2fa015ace061da38b5a) --- ctdb/doc/ctdbd.1.xml | 23 +++++++++++++++++++++++ 1 file changed, 23 insertions(+) diff --git a/ctdb/doc/ctdbd.1.xml b/ctdb/doc/ctdbd.1.xml index 589a8247f0..2dc723d55a 100644 --- a/ctdb/doc/ctdbd.1.xml +++ b/ctdb/doc/ctdbd.1.xml @@ -963,6 +963,29 @@ + FetchCollapse + Default: 1 + + When many clients across many nodes try to access the same record at the + same time this can lead to a fetch storm where the record becomes very + active and bounces between nodes very fast. This leads to high CPU + utilization of the ctdbd daemon, trying to bounce that record around + very fast, and poor performance. + + + This parameter is used to activate a fetch-collapse. A fetch-collapse + is when we track which records we have requests in flight so that we only + keep one request in flight from a certain node, even if multiple smbd + processes are attemtping to fetch the record at the same time. This + can improve performance and reduce CPU utilization for certain + workloads. + + + This timeout controls if we should collapse multiple fetch operations + of the same record into a single request and defer all duplicates or not. + + + LVS -- cgit