Thrift server stops responding - OpenSSL BIO_get_fd error

32 Views Asked by At

I've a thrift server process running off of OpenSSL which starts producing the error "TConnectedClient died: BIO_get_fd failed" message. The server functions well for. afew days and then starts producing the above error. Once the message starts appearing the server stops responding to any client request. Restarting the server process does not help. However, rebooting the entire system appears to help. Once the server process stops responding, I checked lsof output of the server process to see if we're out of file descriptors and things look good there. OpenSSL version is 1.1.1 and thrift version is 0.12

Some additional ideas appreciated on the following

  1. If it is a resource allocation issue, why wouldn't the process restart help that?
  2. Does OpenSSL store global state information anywhere in the system? This might help figure out if there's a resource leak and what kind of resources are leaked.
  3. Next step to debug BIO_get_fd error (assuming the thrift client usage is correct)?
  4. Can there be a corruption of OpenSSL state? Again, why wouldn't restarting the offending process fix that.
0

There are 0 best solutions below