Automatically test various TransferPlugin failure modes
On feature/23111_BrokenTransferTest, I have implemented a test program which tests several possible Transfer failure modes (sender pause, sender reconnect, receiver pause, receiver reconnect), and ensures that the transfer plugins behave correctly (resume/reconnect) in each of these situations.
#3 Updated by Ron Rechenmacher 3 months ago
- % Done changed from 0 to 100
- Status changed from Resolved to Reviewed
all tests pass and verified that SIGPIPE does not cause a crash.
/home/ron/work/artdaqPrj/demo2-develop/srcs/artdaq/test/TransferPlugins/fcl ron@mu2edaq13 :^) treset; time broken_transfer_driver -c broken_transfer_driver_tcp.fcl; echo \$?=$?; tshow | grep ERROR real 1m11.512s user 0m0.125s sys 0m0.236s $?=0 225 1571528343794095 209816 209858 39 _TCPSocketTransfer wrn . transfer_between_1_and_2_SEND: sendFragment_: WRITE ERROR 104: Connection reset by peer 247 1571528343293701 209816 209857 25 _TCPSocketTransfer wrn . transfer_between_0_and_2_SEND: sendFragment_: WRITE ERROR 32: Broken pipe 253 1571528343293329 209816 209858 39 _TCPSocketTransfer wrn . transfer_between_1_and_2_SEND: sendFragment_: WRITE ERROR 32: Broken pipe --2019-10-19_18:39:10--