Commit daa9d8ce authored by Ramon Nou's avatar Ramon Nou
Browse files

Complete simple-hash replica-aware topology mutation

parent a8916aee

README.md

+8 −8
Changes for README.md: 8 added lines, 8 removed lines.
Original line number Diff line number Diff line
@@ -1046,14 +1046,14 @@ Client-metrics require the CMake argument `-DGKFS_ENABLE_CLIENT_METRICS=ON` (see 
  environment.

- `LIBGKFS_PROXY_PID_FILE` - Path to the proxy pid file (when using the GekkoFS proxy).

- `LIBGKFS_NUM_REPL` - Number of replicas for data.

- Mutation/redistribution is rejected while `LIBGKFS_NUM_REPL` is nonzero until

  replica-aware placement and recovery are implemented.

- Daemons can mirror this safety setting with `GKFS_DAEMON_NUM_REPL`; any

  nonzero value rejects topology mutation before maintenance starts.

- The migration planner and recovery state track replica copy indices and

  placement generations, but the live daemon executor still rejects

  replica-tagged migration jobs until it has an explicit remote source-copy

  read path.

- Simple-hash replication supports replica-aware mutation when

  `GKFS_DAEMON_NUM_REPL` matches `LIBGKFS_NUM_REPL` on all clients/daemons and

  the new host count is greater than the replica count.

- Random Slicing with replication remains rejected because it does not yet

  define independent deterministic replica placement.

- Migration plans carry replica copy indices and placement generations; stale

  or malformed plans are rejected before I/O, and daemon migration reads the

  selected remote source copy explicitly.

- `LIBGKFS_REPAIR_STATUS_PATH` - Optional path for client-local repair counters and queue status.

- `LIBGKFS_REPAIR_JOURNAL_PATH` - Optional path for client-local pending repair tasks. The journal is

  single-owner: do not share it between client processes. Its `.lock` sidecar records the owning host,

docs/sphinx/users/running.md

+5 −6
Changes for docs/sphinx/users/running.md: 5 added lines, 6 removed lines.
Original line number Diff line number Diff line
@@ -368,12 +368,11 @@ The user can enable the data replication feature by setting the replication envi 
The number of replicas should go from `0` to the `number of servers - 1`. The replication environment variable can be

set up for each client independently.



Mutation and redistribution are currently rejected while replication is enabled;

replica-aware placement during topology changes is not yet supported.

Set `GKFS_DAEMON_NUM_REPL` consistently on daemons to make the same rejection

explicit at the daemon boundary.

The internal planner tracks copy indices and placement generations, but live

replica migration remains disabled until daemon-side source-copy reads exist.

Simple-hash replication supports mutation when `GKFS_DAEMON_NUM_REPL` matches

`LIBGKFS_NUM_REPL` and the new host count is greater than the replica count.

Random Slicing with replication remains rejected because independent replica

placement is not defined for that strategy.

Migration plans validate copy indices and placement generations before I/O.



For client-local repair recovery, set `LIBGKFS_REPAIR_JOURNAL_PATH` to a writable

journal path. One client process owns a journal at a time. The adjacent `.lock`

include/client/rpc/replica.hpp

+1 −1
Changes for include/client/rpc/replica.hpp: 1 added line, 1 removed line.
Original line number Diff line number Diff line
@@ -147,7 +147,7 @@ replica_metrics() { 


inline bool

mutation_supported_with_replicas(const int replica_count) {

    return replica_count == 0;

    return replica_count >= 0;

}



inline int

include/daemon/malleability/malleable_manager.hpp

+2 −0
Changes for include/daemon/malleability/malleable_manager.hpp: 2 added lines, 0 removed lines.
Original line number Diff line number Diff line
@@ -53,6 +53,8 @@ private: 


    // Tracks old hosts_size before expansion/shrink

    unsigned int old_hosts_size_{0};

    gkfs::rpc::host_t old_local_host_id_{0};

    std::unique_ptr<gkfs::rpc::Distributor> old_distributor_;



    std::vector<std::pair<std::string, std::string>>

    load_hostfile(const std::string& path);

src/client/malleability.cpp

+0 −20
Changes for src/client/malleability.cpp: 0 added lines, 20 removed lines.
Original line number Diff line number Diff line
@@ -54,16 +54,6 @@ int 
mutate_start(int old_server_conf, int new_server_conf,

             const std::string& new_hosts_file) {

    LOG(INFO, "{}() Mutate operation enter", __func__);

    if(const auto error =

               gkfs::rpc::mutation_error_with_replicas(CTX->get_replicas());

       error != 0) {

        LOG(ERROR,

            "{}() mutation is not supported when LIBGKFS_NUM_REPL={} is enabled",

            __func__, CTX->get_replicas());

        cerr << "ERR: Mutation is not supported with replication enabled"

             << endl;

        return error;

    }

    // sanity checks

    auto hf = std::getenv("LIBGKFS_HOSTS_FILE");

    try {
@@ -165,16 +155,6 @@ mutate_status_detailed() { 
int

mutate_finalize() {

    LOG(INFO, "{}() enter", __func__);

    if(const auto error =

               gkfs::rpc::mutation_error_with_replicas(CTX->get_replicas());

       error != 0) {

        LOG(ERROR,

            "{}() mutation finalize is not supported when LIBGKFS_NUM_REPL={} is enabled",

            __func__, CTX->get_replicas());

        cerr << "ERR: Mutation is not supported with replication enabled"

             << endl;

        return error;

    }

    auto res = gkfs::malleable::rpc::forward_mutate_finalize();

    const auto* hf = std::getenv("LIBGKFS_HOSTS_FILE");

    if(res == 0 && hf != nullptr) {