You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Previously, the c2c/disconnect roachtest was disconnecting to source nodes,
instead of a pair of source/destination nodes. This patch fixes this bug.
Informs: #152248
Informs: #150474
Release note: none
srcTenantSQL.QueryRow(t, `select split_part(consumer, '[', 1) from crdb_internal.cluster_replication_node_streams order by random() limit 1`).Scan(&dstNode)
2059
2059
2060
+
roachprodDstNode:=dstNode+sp.srcNodes
2061
+
2060
2062
disconnectDuration:=sp.additionalDuration
2061
2063
rd.t.L().Printf("Disconnecting Src %d, Dest %d for %.2f minutes", srcNode,
2062
-
dstNode, disconnectDuration.Minutes())
2064
+
roachprodDstNode, disconnectDuration.Minutes())
2063
2065
2064
2066
// Normally, the blackholeFailer is accessed through the failer interface,
2065
2067
// at least in the failover tests. Because this test shouldn't use all the
2066
2068
// failer interface calls (e.g. Setup(), and Ready()), we use the
2067
2069
// blakholeFailer struct directly. In other words, in this test, we
2068
2070
// shouldn't treat the blackholeFailer as an abstracted api.
0 commit comments