If the mirror attempt connects ok, but is rejected (say, for reporting
the wrong size), the client socket needs to be closed. The destination
end can't close its socket and accept another connection attempt unless
it does.
Without this, the error you get is a "Bad magic", when the next read
loop tries to read write data as a request. This should be flushed from
the socket (although *when* is an open question), but upping the log
level at least gives us a more informative output.
Previously, the --verbose flag was only present in debug builds. Now
it's present whether you define DEBUG or not. What changes is the
amount of information printed to stderr: DEBUG sets the --verbose log
level to 0 (debug), while DEBUG unset sets it to 1 (info). This makes
driving the binary slightly simpler as you don't have to detect whether
it's a debug build by scanning for "--verbose" in the help output.
If the client cuts off part-way through the write, it should cause an
error, not a fatal. Previously this happened if the open file had a
fiemap, but not if there was no allocation map. This patch fixes that,
along with an associated valgrind error.
O_DIRECT causes problems on (at least) a wheezy VM, and there are mixed
reports about its performance impact. This patch makes it a
compile-time choice which should remain until it's been benchmarked.
Previously, the behaviour was to unlink any control socket sat where we
wanted to open ours. This would make us lose control of running servers
if we happened to collide accidentally. With this patch, the new
process will abort() if there is a control socket squatting on the
path we want, and unlink it when it closes.
This means that an unclean shutdown will leave a dangling, unattached
control socket which will block a restart, but that's a better option
than intentionally cutting off running servers.
This simplifies building the log output because it means we don't have
to malloc a buffer to append a newline, and we keep the atomic write
property we're after. It also takes advantage of the C constant string
concatenation which we already require to work to prepend the thread and
pid data.
When we receive a migration, if rebinding to the new listen address and
port fails for a reason which might be fixable, rather than killing the
server we retry once a second. Also in this patch: non-overlapping log
messages and a fix for the client going away halfway through a sendfile
loop.
If the sender disconnects its socket before sending the disconnect
message, the destination should restart the migration process. This
patch makes sure that happens.
This adds a test for destination behaviour, in that if a source crashes
after sending an entrust message but before the destination can reply,
the destination must allow the source to reconnect and retry the mirror.
Now that we have 3 mutexes lying around, it's important that we check
and free these if necessary if error() is called in any thread that can
hold them. To do this, we now have flexthread.c, which defines a
flexthread_mutex struct. This is a wrapper around a pthread_mutex_t and
a pthread_t. The idea is that in the error handler, the thread can
check whether it holds the mutex and can free it if and only if it does.
This is important because pthread fast mutexes can be freed by *any*
thread, not just the thread which holds them.
Note: it is only ever safe for a thread to check if it holds the mutex
itself. It is *never* safe to check if another thread holds a mutex
without first locking that mutex, which makes the whole operation rather
pointless.
