On 2015-12-26 17:11, Guillermo wrote:
I guess that for a short window of time, processes that send their logs to the catch-all logger could potentially receive a SIGPIPE, right?
Yes, if they fail to ignore SIGPIPE, but most long-lived processes rightfully ignore it. Most likely, a write() would fail with -1 EPIPE.
But would the failing write operations to the logger's FIFO make this a (transient) situation where logs could be lost?
Yes, it would indeed. The solution is to open the named pipe for reading and store the fd into a fd-holder. But it's something for an admin or a distrib to do, it's outside the scope of s6, s6-linux-init or even s6-rc (though I could update the s6-rc example database so it includes that). -- Laurent
