Question

I’m running a command similar to the following

getHosts | xargs -I{} -P3 -n1 ssh {} 'startServer; sleep 5; grep -m 1 "server up" <(tail -f log)'

The problem is that it seems like ssh hangs for a while sometimes even well after the server has come up. Is there any problem with this command that might cause it not to terminate so that parallel execution can continue? When I run the command in a remote shell, the check for the server coming up seems reliable and closes punctually when “server up” is written to the logs.

Answer 1

Instead of the remote command being

startServer; sleep 5; grep -m 1 "server up" <(tail -f log)

I’d use

grep -m 1 "server up" <(tail -F log -n 0) & startServer ; wait

Differences:

• Start tailing the log before attempting to restart the server, so that we don’t miss any messages. We start at the end of the log so we don’t see any previous “server up” messages.
• Use tail‘s -F option instead of -f, so that if the log file is rotated we will follow the new file, instead of continuing to uselessly follow the old file.