wbtiger
/
foundationdb
mirror of https://github.com/apple/foundationdb.git
1
0
Fork 0
Code Issues Packages Projects Releases Wiki Activity

ServerTeamRemover:Believe all servers are healthy when we start to remove 

Before the serverTeamRemover tries to pick a team to remove,
it waits for all data movement to finish, which means all teams are healthy.

When the serverTeamRemover starts to pick a team to remove,
we believe all servers are healthy.
This commit is contained in:
Meng Xu 2019-07-11 23:47:29 -07:00
parent cf935ff9e6
commit 4da2071b49
1 changed files with 1 additions and 12 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
fdbserver/DataDistribution.actor.cpp
View File

@ -2570,19 +2570,9 @@ ACTOR Future<Void> serverTeamRemover(DDTeamCollection* self) {
// adding the bad team (add the team tracker) and remove bad team (cancel the team tracker).
wait(self->badTeamRemover);
// Q: Should we count the number of servers instead of healthy servers, since healthyness can change quickly?
int healthyServerCount = self->calculateHealthyServerCount();
// Check if all servers are healthy, if not, we wait for 1 second and loop back.
// Eventually, all servers will become healthy.
if (healthyServerCount != self->server_info.size()) {
continue;
}
// From this point, all server teams should be healthy, because we wait above
// until processingUnhealthy is done, and all machines are healthy
// In most cases, all machine teams should be healthy teams at this point.
int desiredServerTeams = SERVER_KNOBS->DESIRED_TEAMS_PER_SERVER * healthyServerCount;
int desiredServerTeams = SERVER_KNOBS->DESIRED_TEAMS_PER_SERVER * self->server_info.size();
int totalSTCount = self->teams.size();
if (totalSTCount > desiredServerTeams) {
@ -2616,7 +2606,6 @@ ACTOR Future<Void> serverTeamRemover(DDTeamCollection* self) {
if (numServerTeamRemoved > 0) {
// Only trace the information when we remove a machine team
TraceEvent("ServerTeamRemoverDone", self->distributorId)
.detail("HealthyServerNumber", healthyServerCount)
.detail("CurrentServerTeamNumber", self->teams.size())
.detail("DesiredServerTeam", desiredServerTeams)
.detail("NumServerTeamRemoved", numServerTeamRemoved);