## Problem
We have been dealing with a number of issues with the SC compute
notification mechanism. Various race conditions exist in the
PG/HCC/cplane/PS distributed system, and relying on the SC to send
notifications to the compute node to notify it of PS changes is not
robust. We decided to pursue a more robust option where the compute node
itself discovers whether it may be pointing to the incorrect PSs and
proactively reconfigure itself if issues are suspected.
## Summary of changes
To support this self-healing reconfiguration mechanism several pieces
are needed. This PR adds a mechanism to `compute_ctl` called "refresh
configuration", where the compute node reaches out to the control plane
to pull a new config and reconfigure PG using the new config, instead of
listening for a notification message containing a config to arrive from
the control plane. Main changes to compute_ctl:
1. The `compute_ctl` state machine now has a new State,
`RefreshConfigurationPending`. The compute node may enter this state
upon receiving a signal that it may be using the incorrect page servers.
2. Upon entering the `RefreshConfigurationPending` state, the background
configurator thread in `compute_ctl` wakes up, pulls a new config from
the control plane, and reconfigures PG (with `pg_ctl reload`) according
to the new config.
3. The compute node may enter the new `RefreshConfigurationPending`
state from `Running` or `Failed` states. If the configurator managed to
configure the compute node successfully, it will enter the `Running`
state, otherwise, it stays in `RefreshConfigurationPending` and the
configurator thread will wait for the next notification if an incorrect
config is still suspected.
4. Added various plumbing in `compute_ctl` data structures to allow the
configurator thread to perform the config fetch.
The "incorrect config suspected" notification is delivered using a HTTP
endpoint, `/refresh_configuration`, on `compute_ctl`. This endpoint is
currently not called by anyone other than the tests. In a follow up PR I
will set up some code in the PG extension/libpagestore to call this HTTP
endpoint whenever PG suspects that it is pointing to the wrong page
servers.
## How is this tested?
Modified `test_runner/regress/test_change_pageserver.py` to add a
scenario where we use the new `/refresh_configuration` mechanism instead
of the existing `/configure` mechanism (which requires us sending a full
config to compute_ctl) to have the compute node reload and reconfigure
its pageservers.
I took one shortcut to reduce the scope of this change when it comes to
testing: the compute node uses a local config file instead of pulling a
config over the network from the HCC. This simplifies the test setup in
the following ways:
* The existing test framework is set up to use local config files for
compute nodes only, so it's convenient if I just stick with it.
* The HCC today generates a compute config with production settings
(e.g., assuming 4 CPUs, 16GB RAM, with local file caches), which is
probably not suitable in tests. We may need to add another test-only
endpoint config to the control plane to make this work.
The config-fetch part of the code is relatively straightforward (and
well-covered in both production and the KIND test) so it is probably
fine to replace it with loading from the local config file for these
integration tests.
In addition to making sure that the tests pass, I also manually
inspected the logs to make sure that the compute node is indeed
reloading the config using the new mechanism instead of going down the
old `/configure` path (it turns out the test has bugs which causes
compute `/configure` messages to be sent despite the test intending to
disable/blackhole them).
```test
2024-09-24T18:53:29.573650Z INFO http request{otel.name=/refresh_configuration http.method=POST}: serving /refresh_configuration POST request
2024-09-24T18:53:29.573689Z INFO configurator_main_loop: compute node suspects its configuration is out of date, now refreshing configuration
2024-09-24T18:53:29.573706Z INFO configurator_main_loop: reloading config.json from path: /workspaces/hadron/test_output/test_change_pageserver_using_refresh[release-pg16]/repo/endpoints/ep-1/spec.json
PG:2024-09-24 18:53:29.574 GMT [52799] LOG: received SIGHUP, reloading configuration files
PG:2024-09-24 18:53:29.575 GMT [52799] LOG: parameter "neon.extension_server_port" cannot be changed without restarting the server
PG:2024-09-24 18:53:29.575 GMT [52799] LOG: parameter "neon.pageserver_connstring" changed to "postgresql://no_user@localhost:15008"
...
```
Co-authored-by: William Huang <william.huang@databricks.com>
4.3 KiB
Compute node tools
Postgres wrapper (compute_ctl) is intended to be run as a Docker entrypoint or as a systemd
ExecStart option. It will handle all the Neon specifics during compute node
initialization:
compute_ctlaccepts cluster (compute node) specification as a JSON file.- Every start is a fresh start, so the data directory is removed and initialized again on each run.
- Next it will put configuration files into the
PGDATAdirectory. - Sync safekeepers and get commit LSN.
- Get
basebackupfrom pageserver using the returned on the previous step LSN. - Try to start
postgresand wait until it is ready to accept connections. - Check and alter/drop/create roles and databases.
- Hang waiting on the
postmasterprocess to exit.
Also compute_ctl spawns two separate service threads:
compute-monitorchecks the last Postgres activity timestamp and saves it into the sharedComputeNode;http-endpointruns a Hyper HTTP API server, which serves readiness and the last activity requests.
If AUTOSCALING environment variable is set, compute_ctl will start the
vm-monitor located in [neon/libs/vm_monitor]. For VM compute nodes,
vm-monitor communicates with the VM autoscaling system. It coordinates
downscaling and requests immediate upscaling under resource pressure.
Usage example:
compute_ctl -D /var/db/postgres/compute \
-C 'postgresql://cloud_admin@localhost/postgres' \
-S /var/db/postgres/specs/current.json \
-b /usr/local/bin/postgres
State Diagram
Computes can be in various states. Below is a diagram that details how a compute moves between states.
%% https://mermaid.js.org/syntax/stateDiagram.html
stateDiagram-v2
[*] --> Empty : Compute spawned
Empty --> ConfigurationPending : Waiting for compute spec
ConfigurationPending --> Configuration : Received compute spec
Configuration --> Failed : Failed to configure the compute
Configuration --> Running : Compute has been configured
Empty --> Init : Compute spec is immediately available
Empty --> TerminationPendingFast : Requested termination
Empty --> TerminationPendingImmediate : Requested termination
Init --> Failed : Failed to start Postgres
Init --> Running : Started Postgres
Running --> TerminationPendingFast : Requested termination
Running --> TerminationPendingImmediate : Requested termination
Running --> ConfigurationPending : Received a /configure request with spec
Running --> RefreshConfigurationPending : Received a /refresh_configuration request, compute node will pull a new spec and reconfigure
RefreshConfigurationPending --> Running : Compute has been re-configured
TerminationPendingFast --> Terminated compute with 30s delay for cplane to inspect status
TerminationPendingImmediate --> Terminated : Terminated compute immediately
Running --> TerminationPending : Requested termination
TerminationPending --> Terminated : Terminated compute
Failed --> RefreshConfigurationPending : Received a /refresh_configuration request
Failed --> [*] : Compute exited
Terminated --> [*] : Compute exited
Tests
Cargo formatter:
cargo fmt
Run tests:
cargo test
Clippy linter:
cargo clippy --all --all-targets -- -Dwarnings -Drust-2018-idioms
Cross-platform compilation
Imaging that you are on macOS (x86) and you want a Linux GNU (x86_64-unknown-linux-gnu platform in rust terminology) executable.
Using docker
You can use a throw-away Docker container (rustlang/rust image) for doing that:
docker run --rm \
-v $(pwd):/compute_tools \
-w /compute_tools \
-t rustlang/rust:nightly cargo build --release --target=x86_64-unknown-linux-gnu
or one-line:
docker run --rm -v $(pwd):/compute_tools -w /compute_tools -t rust:latest cargo build --release --target=x86_64-unknown-linux-gnu
Using rust native cross-compilation
Another way is to add x86_64-unknown-linux-gnu target on your host system:
rustup target add x86_64-unknown-linux-gnu
Install macOS cross-compiler toolchain:
brew tap SergioBenitez/osxct
brew install x86_64-unknown-linux-gnu
And finally run cargo build:
CARGO_TARGET_X86_64_UNKNOWN_LINUX_GNU_LINKER=x86_64-unknown-linux-gnu-gcc cargo build --target=x86_64-unknown-linux-gnu --release