----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/23147/#review46964 -----------------------------------------------------------
Looks great, Alexandra! Much more readable now. I know it's a lot of code to go through, so I appreciate your effort. There are still a couple more related variables/methods left to rename (Master::Slave::disconnected, Master::disconnect(Slave*), InvokeSlaveReconnected()), and some documentation updates. Once that is complete, this will be ready to commit. Thanks! src/master/allocator.hpp <https://reviews.apache.org/r/23147/#comment82538> s/reconnected/reactivated/ src/master/hierarchical_allocator_process.hpp <https://reviews.apache.org/r/23147/#comment82539> s/until they reactivated/until they are reactivated/ src/master/http.cpp <https://reviews.apache.org/r/23147/#comment82541> There could be upgrade issues for anyone using hardcoded keys to read from the stats endpoint. I'm not sure how important that is, but I can think of two approaches: 1) Just document the change. 2) Document the change, and store these stats at the old keys as well as the new ones. The old keys can be deprecated in a future release. src/master/http.cpp <https://reviews.apache.org/r/23147/#comment82540> I wonder if "total_schedulers" is the correct term for this. Perhaps "registered_schedulers" would be more explicit? src/master/http.cpp <https://reviews.apache.org/r/23147/#comment82542> BTW, why are these "schedulers" and not "frameworks"? Should we rename them while we're at it? src/master/http.cpp <https://reviews.apache.org/r/23147/#comment82543> Again, we'll need to document the change and possibly deprecate the old keys. src/master/master.hpp <https://reviews.apache.org/r/23147/#comment82558> Remove this TODO src/master/master.hpp <https://reviews.apache.org/r/23147/#comment82559> deactivate src/master/master.hpp <https://reviews.apache.org/r/23147/#comment82560> // A registered slave. Or just remove the comment. It doesn't really add much. src/master/master.hpp <https://reviews.apache.org/r/23147/#comment82561> active(true), src/master/master.hpp <https://reviews.apache.org/r/23147/#comment82562> // We mark a slave 'inactive' ... bool active; src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82545> } else if (slave->active) { // Checkpointing slaves can just be deactivated. deactivate(slave); src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82546> deactivate(Slave*) src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82547> Deactivating src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82548> deactivated src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82549> slave->active = false; src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82550> if (!slave->active) { src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82551> deactivated src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82552> deactivated src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82553> // If this is a deactivated slave... if (!slave->active) { slave->active = true; src/master/master.cpp <https://reviews.apache.org/r/23147/#comment82554> I wonder if we should also be checking if (!slaves.registered[slaveId]->active) src/tests/mesos.hpp <https://reviews.apache.org/r/23147/#comment82556> InvokeSlaveReactivated src/tests/mesos.hpp <https://reviews.apache.org/r/23147/#comment82557> InvokeSlaveReactivated - Adam B On June 28, 2014, 9:19 a.m., Alexandra Sava wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/23147/ > ----------------------------------------------------------- > > (Updated June 28, 2014, 9:19 a.m.) > > > Review request for mesos, Adam B and Ben Mahler. > > > Bugs: MESOS-1188 > https://issues.apache.org/jira/browse/MESOS-1188 > > > Repository: mesos-git > > > Description > ------- > > The existing terminology is confusing both for "slaves.deactivated" and > "frameworks.activated". Currently a deactivated slave actually represents a > removed/shutdown slave and "frameworks.activated" map holds both activated > and deactivated frameworks. > In order to make things look clear, rename the following: > * master.slaves.deactivated -> master.slaves.removed > * master.slaves.activated -> master.slaves.registered > * master.frameworks.activated -> master.frameworks.registered > * allocator.slaveDisconnect -> allocator.slaveDeactivate > * allocator.slaveReconnected -> allocator.slaveReactivated > > > Diffs > ----- > > src/master/allocator.hpp 1cd573477b609bb69264f16156a4004ecac672a7 > src/master/constants.hpp 2daa9b004ab0cc79773730350369f66315356cad > src/master/constants.cpp e9e5e67f890f3399c24637c0f021d656dfe51118 > src/master/hierarchical_allocator_process.hpp > 1765e7035bdda4c28e79d74c92e77dcc99759001 > src/master/http.cpp 5d869767cd3ed48aae1e702e8d014a37ef371123 > src/master/master.hpp 5fef35406c2ce2ad11e030aa7752eb691aab5857 > src/master/master.cpp 21b07c7f1f445beac29a7781cf441dd79b1b7fb5 > src/tests/fault_tolerance_tests.cpp > ac65050bec5720b982f53d4dd6797cc3dee285dc > src/tests/master_authorization_tests.cpp > 478041cdea533e548ca92c4b8e8c793554855969 > src/tests/mesos.hpp ae38a13d8b329f6e27813776e0d2f2b56605d0eb > src/tests/slave_recovery_tests.cpp 582f52d73eba0e3ab089ec573d9a6c43bff0339e > src/webui/master/static/home.html ce8ca192235c224715c01fef0b8ddb187dc0a827 > src/webui/master/static/js/controllers.js > 41a70a80442501a2bf7b217939dbe504662941d2 > > Diff: https://reviews.apache.org/r/23147/diff/ > > > Testing > ------- > > > Thanks, > > Alexandra Sava > >
