apple/foundationdb
1
0
Fork 0
mirror of https://github.com/apple/foundationdb.git synced 2025-06-02 03:12:12 +08:00
Code Issues Packages Projects Releases Wiki Activity

Changed logic and reordered swap of coordinator exclusion in workload

Browse Source
This commit is contained in:
Jon Fu 2019-10-09 10:22:42 -07:00
parent eb41e32876
commit ac7369d27c
1 changed files with 13 additions and 9 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

22
fdbserver/workloads/RemoveServersSafely.actor.cpp
View File

@ -407,19 +407,11 @@ struct RemoveServersSafelyWorkload : TestWorkload {
if (BUGGIFY && toKill.size()) {
std::vector<NetworkAddress> coordinators = wait(getCoordinators(cx));
if (coordinators.size() > 2) {
auto removeServer = toKill.begin();
auto randomCoordinator = deterministicRandom()->randomChoice(coordinators);
coordExcl = AddressExclusion(randomCoordinator.ip, randomCoordinator.port);
TraceEvent("RemoveAndKill", functionId)
.detail("Step", "ReplaceKillSet")
.detail("Removing", removeServer->toString())
.detail("Adding", coordExcl.toString());
toKill.erase(removeServer);
toKill.insert(coordExcl);
}
}
std::copy(toKill.begin(), toKill.end(), std::back_inserter(toKillArray));
killProcArray = self->getProcesses(toKill);
if (markExcludeAsFailed) {
state int retries = 0;
loop {
@ -456,7 +448,19 @@ struct RemoveServersSafelyWorkload : TestWorkload {
retries++;
}
}
// Swap out coordinator with server in kill set, but only if already marking as failed and safety check passes
if (markExcludeAsFailed && coordExcl.isValid()) {
auto removeServer = toKill.begin();
TraceEvent("RemoveAndKill", functionId)
.detail("Step", "ReplaceKillSet")
.detail("Removing", removeServer->toString())
.detail("Adding", coordExcl.toString());
toKill.erase(removeServer);
toKill.insert(coordExcl);
toKillArray.erase(std::remove(toKillArray.begin(), toKillArray.end(), *removeServer), toKillArray.end());
toKillArray.push_back(coordExcl);
}
killProcArray = self->getProcesses(toKill);
TraceEvent("RemoveAndKill", functionId).detail("Step", "Activate Server Exclusion").detail("KillAddrs", toKill.size()).detail("KillProcs", killProcArray.size()).detail("MissingProcs", toKill.size()!=killProcArray.size()).detail("ToKill", describe(toKill)).detail("Addresses", describe(toKillArray)).detail("ClusterAvailable", g_simulator.isAvailable());
if (markExcludeAsFailed) {
wait( excludeServers( cx, toKillMarkFailedArray, true ) );