clsync ignores SIGTERM

[bircoph]

Minimal sample:

/usr/bin/clsync --dir-lists /tmp/clsync -RR /root/scripts /usr/bin/rsync /etc/clsync/rules /root/test

/root/scripts/: just some files with several dirs
/etc/clsync/rules:
+.

Ignores SIGTERM, dies only on SIGKILL.

[xaionaro]

How long did you wait? Also can you show output of "du -ms /root/scripts", "find /root/scripts -type f | wc -l" and "find /root/scripts | wc -l"?

[bircoph]

For half an hour. -w2 -t2 -p options doesn't affect this.

du -ms /root/scripts/
2 /root/scripts/

find /root/scripts -type f | wc -l
244

find /root/scripts | wc -l
376

[bircoph]

SIGINT is also ignored. Looks like some bug in the signal processing code.

[bircoph]

Output of:
strace -ff -o clsync.strace /usr/bin/clsync --pid-file=/var/run/clsync.pid --dir-lists /tmp/clsync -RR -w2 -t2 /root/scripts /usr/bin/rsync /etc/clsync/rules /root/test
is here: ftp://arcueid.brunestud.info/clsync.strace.tar.xz

[xaionaro]

Cannot repeat the bug. Running with the same options and it works :(

Hm. Are you able to give me the "/root/scripts" dir? You can encrypt it with PGP. My PGP key is 0x8E30679C, as you already know, :).

If you're unable to provide me with the "scripts" dir, can you run clsync with "-DDD" option and copy the log somewhere (e.g. to http://pb.campus.mephi.ru/)?

[xaionaro]

Fixed in: afe5e83

[xaionaro]

I forgot about a bug in that part of code. I was getting a infinite wait if tv_sec is incremented. So I was have to temporary remove the increment. But I also forgot to comment about this problem in the code and in the TODO.

But now there's no problem with increment, and it's required to make the code work well... so it's returned.

[bircoph]

It's not fully fixed, now we have a segfault on exit.

[xaionaro]

It's another issue :)

[bircoph]

Do you want me to open another bug?

[xaionaro]

If you feel like it :).

I just don't have enough information to open the issue at the moment. I'll continue to work on clsync tomorrow.

[bircoph]

Strange the latest clsync doesn't segfault on exit anymore. This is good, but is unclear why.