I'm running a plain 2.1.25 (no patches) on a Red Hat 4.0 system.
The problem is that "dump" freezes at the _end_ of the backup operation.
With 2.1.24, everything was ok.
I compared 2.0.27 and 2.1.25 (both Red Hat 4.0) with "strace" and
observerd the following:
dump forks a second control process (which does stdout) and three more
processes which do the backup work in parallel. The problem seems to be
in these three "work" processes. Note the 264-byte read:
2.0.27:
=======
lseek(13, 47813632, SEEK_SET) = 47813632
read(13, "\nread(12, \"\\2\\210\\0\\0\\f\\"..., 1024) = 1024
sigprocmask(SIG_BLOCK, NULL, []) = 0
write(14, " 1024\nlseek(12, 168039424, SEEK"..., 32768) = 32768
write(15, "\0\200\0\0", 4) = 4
kill(5270, SIGUSR2) = 0
read(15, "", 264) = 0
_exit(0) = ?
2.1.25:
=======
read(19, "\5\0\0\0\23E\3702\2559\3702\1\0\0"..., 1024) = 1024
read(19, "\5\0\0\0\23E\3702\2559\3702\1\0\0"..., 1024) = 1024
sigprocmask(SIG_BLOCK, NULL, []) = 0
write(18, "rver 44.143.22.1 offset 0.065277"..., 32768) = 32768
write(19, "\0\200\0\0", 4) = 4
kill(6459, SIGUSR2) = 0
read(19, 0x805def8, 264) = ? ERESTARTSYS (To be restarted)
--- SIGTERM (Terminated) ---
getpid() = 6458
kill(6457, SIGTERM) = 0
_exit(3) = ?
Do you see the difference? :-)
I had to manually kill the shown "work" process, since the 264-byte read
blocked (actually all three "work" processes were blocked by such a
264-byte read).
The same read returns zero on 2.0.27 and everything is ok....
Any idea?
-Klaus
-- Klaus Kudielka OE1KIB Peter Jordanstr. 165, A-1180 Wien, AUSTRIA oe1kib@oe1xtu.ampr.org http://oe1xtu.ampr.org/~oe1kib/ OE1KIB@OE1XIB.AUT.EU http://asterix.nt.tuwien.ac.at/~oe1kib/