Fix a major memory issue in the SumStats framework.

- There are still problems, but this should prevent a deadlock issue
   and help with memory use.
This commit is contained in:
Seth Hall 2013-08-03 01:57:37 -04:00
parent 56de65461e
commit 1eadeaec3c
3 changed files with 7 additions and 3 deletions

View file

@ -1,4 +1,8 @@
2.1-1027 | 2013-08-03 01:57:37 -0400
* Fix a major memory issue in the SumStats framework.
2.1-1026 | 2013-08-02 22:35:09 -0400
* Fix the SumStats top-k plugin and test. (Seth Hall)

View file

@ -1 +1 @@
2.1-1026
2.1-1027

View file

@ -73,7 +73,7 @@ global recent_global_view_keys: table[string, Key] of count &create_expire=1min
# Result tables indexed on a uid that are currently being sent to the
# manager.
global sending_results: table[string] of ResultTable = table();
global sending_results: table[string] of ResultTable = table() &create_expire=1min;
# This is done on all non-manager node types in the event that a sumstat is
# being collected somewhere other than a worker.
@ -171,7 +171,7 @@ event SumStats::cluster_ss_request(uid: string, ss_name: string, cleanup: bool)
#print fmt("WORKER %s: received the cluster_ss_request event for %s.", Cluster::node, id);
# Create a back store for the result
sending_results[uid] = (ss_name in result_store) ? copy(result_store[ss_name]) : table();
sending_results[uid] = (ss_name in result_store) ? result_store[ss_name] : table();
# Lookup the actual sumstats and reset it, the reference to the data
# currently stored will be maintained internally from the