You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have checked the network connection between worker and miner, The network connection is no problem
# in our worker
telnet 10.0.8.101 2345
# output
Trying 10.0.8.101...
Connected to 10.0.8.101.
Escape character is '^]'.
^C
Connection closed by foreign host.
Then I realized that the TCP-Backlog queue may be full
# in our miner
netstat -s | grep LISTEN
193 SYNs to LISTEN sockets dropped
At the same time, I also found that there will be a large number of TIME_WAIT
netstat -na |grep TIME_WAIT | wc -l
89542
I review the code of go-jsonrpc about httpClient, I found that connection pool is not used, Why not use go http Transport, I think it should be able to improve the problem of excessive http connections
The text was updated successfully, but these errors were encountered:
Describe the bug
In our big cluster, we can often find in the log that the worker cannot connect to the miner
We have checked the network connection between worker and miner, The network connection is no problem
Then I realized that the TCP-Backlog queue may be full
At the same time, I also found that there will be a large number of
TIME_WAIT
I review the code of go-jsonrpc about httpClient, I found that connection pool is not used, Why not use go http
Transport
, I think it should be able to improve the problem of excessive http connectionsThe text was updated successfully, but these errors were encountered: