# This is the 1st commit message:
Fix travis
# This is the commit message #2:
[cleanup] remove is_genesis flag
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #3:
[nodetype] add nodetype to replace is_explorer
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #4:
fix beacon sync channel blocking issue
# This is the commit message #5:
use lastMileMux to protect lastMileBlock queue to avoid potential blocking; use sleep instead of ticker
# This is the commit message #6:
Fix the beacon committee check
# This is the commit message #7:
[release] add release action to go_executable_build.sh
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #8:
[node.sh] add -d option
-d download only
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #9:
[node.sh] add -T node_type option
-T node_type support node type (validator/explorer)
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #10:
[node.sh] backward compatible with older harmony binary
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #11:
[node.sh] support -i shard_id option
-i shard_id specify shard_id, this is applicable only to
explorer node
Signed-off-by: Leo Chen <leo@harmony.one>
# This is the commit message #12:
Revisited api
# This is the commit message #13:
Fix rpc integration
# This is the commit message #14:
address some minor issues in comments and code
# This is the commit message #15:
addressed comments on others' buckets
# This is the commit message #16:
Add Global Access to OS Temp Directory Variable and Move DHT Files Into Temp Directory
Valid networks are main (mainnet), beta (betanet), pangaea (Pangaea);
default is main. beta is unsupported yet.
Vary release channel (folder) and bootnodes.
TODO @harmony-ek – support betanet
TODO @harmony-ek – replace hardcoded bootnode with DNS
I removed triggered state because how it was functioning before would cause potentially some conflicts with manual pagerduty resolutions. For example, if a node broke, it would save true into triggered, and then only trigger again if it was set to false. It could only be set to false if the node came back up, which, when the script ran again, false would be saved into triggered. However, if the alert was accidentally resolved and the node was still broken, no alert would come up as triggered would still have true. The save state was mainly to prevent multiple posts, but as there is the dedup key and the program doesn't stall if there is no alert to resolve, there's no real reason to have it.