-
Notifications
You must be signed in to change notification settings - Fork 1.5k
Clients only connect to three proxies to reduce the number of network connections #1909
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Clients only connect to three proxies to reduce the number of network connections #1909
Conversation
fdbclient/MonitorLeader.actor.cpp
Outdated
| when( state ClientDBInfo ni = wait( brokenPromiseToNever( knownLeader->get().get().clientInterface.openDatabase.getReply( req ) ) ) ) { | ||
| TraceEvent("MonitorLeaderForProxiesGotClientInfo", knownLeader->get().get().clientInterface.id()).detail("Proxy0", ni.proxies.size() ? ni.proxies[0].id() : UID()).detail("ClientID", ni.id); | ||
| if(ni.proxies.size() > CLIENT_KNOBS->MAX_CLIENT_PROXY_CONNECTIONS) { | ||
| deterministicRandom()->randomShuffle(ni.proxies); |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Since we are mutating the list of proxies here, monitorMasterProxiesChanged will probably get triggered even when it shouldn't. Don't think it should matter much though, but that may also mean that ACTOR is probably unnecessary as it seems that Future is effectively clientDBInfo->onChange() as we'll shuffle this everytime which would make that ACTOR trigger onMasterProxyChanged() each time.
|
@fdb-build, test macos please |
… and continue connecting to the wrong cluster
…ed to happen when speedUpSimulation is true
…atacenter to be considered dead
…ould have made all the clients talking to that coordinator connect to the same set of proxies optimized the code to avoid re-randomizing the same list of proxies
No description provided.