* Add rate limit to metrics
* Compliance with harmony's metrics specifications
* Improve rate limit metrics info
* minor: remove the redundancy of rpc metrics name
Signed-off-by: Leo Chen <leo@harmony.one>
* [rpc] set default rate limit to 1000
Signed-off-by: Leo Chen <leo@harmony.one>
* [rpc] fix test errors
Signed-off-by: Leo Chen <leo@harmony.one>
* [metrics] init rpc rate limiter
Signed-off-by: Leo Chen <leo@harmony.one>
Co-authored-by: Leo Chen <leo@harmony.one>
* [cmd] added --sync as the switch of stream sync protocol
* [cmd] minor formatting fix and optimization
* [cmd] add sanity fixes for compatible for last v2.0.0 config files
Removed implicit port subtraction(-3000). This is now only done in the migration of old configs. Once old configs
are migrated port will match with actual port used
- Changed update config prompt to default to No
- Moved RosettaPort == 0 check to migration 1.0.4
- Refractored migrateConf to use separate func for the migration logic
- DNS port is substracted 3000 only if it is not eqal current
default(6000)
- Setting default value to ServerPort if field is missing from migrated conf
Fixes regarding config V2.0.0
The following fixes and improvements have been applied
- Added timeout to config update prompt. Prompt logic has been moved to
a separate func
- When updating config file, the original is saved to <config>.backup
- Added semantic version validation for the version found in config. Error is
returned if version is missing, not a valid sematic version or not
found in the migrations map
- Flags related to DNSSync has been moved to dnsSyncFlags flag array.
Also moved tests related to DNSSync flags to separate test func
- Added dns.server and dns.client flags and tests for them
- Added migration tests with dumps of versions 1.0.2, 1.0.3, 1.0.4
Inital commit
Removed lint warning
Fixed names, that have changed in dnsSync
Fixed some lint errors
Some functions were exported although no need to be and some test
variables were also renamed
Fixed 1.0.4 migration
- Tests in config_test.go expected wrong version since it was migrated to
2.0.0
- getDefaultHmyConfigCopy did not get DNSSync defaults
- Key for DNSPort for version prior to 1.0.4 was wrong - DNSSyncPort
Added default for DNSSync when creating default Config
Fixed empty string user input crash
Inputing empty string was causing out of bounds index to be read from
readStr
Fixed flags_test to work with config version 2.0.0
Changed DNSSync fields names
It seems that Enabled was not a good name, considering that actually did
the opposite. Also kept LegacyClient and LegacyServer names
Removed Legacy prefix from dnsSync fields
It seems Legacy prefix is obsolite since moving the fields to dsnSync
Fixes regarding config V2.0.0
The following fixes and improvements have been applied
- Added timeout to config update prompt. Prompt logic has been moved to
a separate func
- When updating config file, the original is saved to <config>.backup
- Added semantic version validation for the version found in config. Error is
returned if version is missing, not a valid sematic version or not
found in the migrations map
- Flags related to DNSSync has been moved to dnsSyncFlags flag array.
Also moved tests related to DNSSync flags to separate test func
- Added dns.server and dns.client flags and tests for them
- Added migration tests with dumps of versions 1.0.2, 1.0.3, 1.0.4
Fix for config_migrations_test
Added default value to serverPort during migration
1. Added some functionality to HostV2 for stream support.
2. Added new module discovery under p2p host to replace the networkInfo service.
3. Make dht datastore badger only active on bootstrap nodes. For normal nodes, badger is disabled by default. It can also be enabled by flags.
Co-authored-by: Rongjian Lan <rongjian.lan@gmail.com>
1. Refactored service manager with cleaner interface.
2. Add prometheus to the service manager.
3. Graceful shutdown of the services (including consensus).
4. Some code refactor regarding consensus graceful shutdown.
Co-authored-by: Rongjian Lan <rongjian.lan@gmail.com>
* [node] graceful shutdown validator and leader
when leader/validator is in commit phase, wait until it finishes the consensus
before shutdown the node.
This is useful for node upgrade to avoid unexpected blockchain state.
Signed-off-by: Leo Chen <leo@harmony.one>
* [node] stop block proposal service on leader
Signed-off-by: Leo Chen <leo@harmony.one>
* [node] warn on block proposal stopped
Signed-off-by: Leo Chen <leo@harmony.one>
Co-authored-by: Rongjian Lan <rongjian.lan@gmail.com>
* Add 3s block time and change of block reward
* fixes
* add new blocks per epoch for testnet
* remove unused code
* fix 2s changing logic
* enable 2s at epoch 73000 at testnet
Peter Chung
Jacky Wang
MathxH Chen
ntsanov
Rongjian Lan
Leo Chen