Details
-
Bug
-
Status: Closed (View Workflow)
-
Minor
-
Resolution: Not a Bug
-
10.3.27
-
None
-
RHEL 8, 2 CPU, 7 GB, SSD storage
Description
We recently saw a strange behaviour with 2 new test galera clusters. When Galera is enabled select version() takes 200ms, when we disable Galera it's like 0.00000 sec (which is normal).
We found out because we had some problems with very slow queries (+200 ms too) on a simple select WITHOUT joins on a table with 600 records. The database is like 500 Mb big.
So my guess the slow select version has something todo with it.
I tested it on some older clusters and there select version() is instant.
This is the galera config (25.3.31-1):
[galera]
wsrep_on = ON
wsrep_causal_reads = ON
wsrep_cluster_address = gcomm://node1,node3
wsrep_cluster_name = -test-db-cluster
wsrep_node_address = node2
wsrep_node_name = "db-node-2"
wsrep_provider = /usr/lib64/galera/libgalera_smm.so
wsrep_provider_options = "gcache.size=10G;gcs.fc_limit=120;gcs.fc_factor=0.8"
wsrep_slave_threads = 2
wsrep_sst_auth = "sstuser:xxxxxxxxx"
wsrep_sst_method = mariabackup