Re: stale connections, keepalive?
On Sun, Aug 28, 2016 at 02:13:34PM +0200, Tomasz Sterna wrote: > W dniu 27.08.2016, sob o godzinie 14∶55 -0400, użytkownik Greg Troxel > napisał: > > should jabberd2 force TCP keepalive on? > I'm not sure whether it is possible. > At least on Linux it is a system-wide setting and requires root to > change. Are you sure? There appear to be some socket options that can be set for each socket: http://www.tldp.org/HOWTO/html_single/TCP-Keepalive-HOWTO/#setsockopt Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org
Re: Loqui IM doesn’t work with the XEP-198 by jabberd2
On Tue, May 26, 2015 at 06:08:20PM +0200, Matěj Cepl wrote: (a follow-up to the issue https://github.com/loqui/im/issues/732#issuecomment-105519240) BTW, I believe this is actually the same as discussed here some time ago http://thread.gmane.org/gmane.network.jabber.jabberd2/1454/focus=1458 Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org
expat 2.1.0 breaks jabberd2
Hi, just upgraded my server from Ubuntu 10.04 to 12.04 and noticed that jabberd2 stopped working. I finally tracked it down to the expat library which now contains a fix for CVE-2012-0876. This fix tries to add some randomisation to expat's own hash tables, but by doing so fiddles with srand and thereby makes the random ids generated by jabberd2 much less random. One way around this is by using the new function XML_SetHashSalt just after creating an XML parser... Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org
Re: jabberd-2.2.15 release
On Mon, Apr 30, 2012 at 05:04:22PM +0200, Tomasz Sterna wrote: Dnia 2012-04-30, pon o godzinie 11:06 +0200, Marcin Mirosław pisze: Would it be possible to have tarball in Downloads section of github ( https://github.com/Jabberd2/jabberd2/downloads )? It would, but why? I see no reason for duplication of the same tarballs. Well, the ones in the Downloads section should ideally already contain the autoconf generated configure script. Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org
Re: Problem with setup jabberd2 to working with ipv4 and ipv6
On Fri, Jan 13, 2012 at 05:14:59PM +0100, Marcin Mirosław wrote: I've found information i should set local/ip to :: (double colon), done. What's the setting of net.ipv6.bindv6only (sysctl net.ipv6.bindv6only)? Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org
Re: Problem with setup jabberd2 to working with ipv4 and ipv6
On Fri, Jan 13, 2012 at 08:46:04PM +0100, Marcin Mirosław wrote: W dniu 2012-01-13 20:24, Christof Meerwald pisze: On Fri, Jan 13, 2012 at 05:14:59PM +0100, Marcin Mirosław wrote: I've found information i should set local/ip to :: (double colon), done. What's the setting of net.ipv6.bindv6only (sysctl net.ipv6.bindv6only)? # cat /proc/sys/net/ipv6/bindv6only 0 Looks like the kernel got a bit more strict recently - with 2.6.32 you can create an AF_INET socket and bind it to an AF_INET6 :: address, but with 3.0 the kernel now complains with EAFNOSUPPORT. I guess as a quick and dirty workaround you might be able to set s2s-origin_ip = NULL; in s2s/main.c, line 126 Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org
[Jabberd2] patch: make max_fds configurable for s2s
Hi, see subject, patch attached. Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org diff -ur jabberd-2.1.16.orig/etc/s2s.xml.dist.in jabberd-2.1.16/etc/s2s.xml.dist.in --- jabberd-2.1.16.orig/etc/s2s.xml.dist.in 2007-08-31 02:24:28.0 +0200 +++ jabberd-2.1.16/etc/s2s.xml.dist.in 2007-09-23 21:30:49.0 +0200 @@ -121,6 +121,16 @@ /local + !-- input/output settings -- + io +!-- Maximum number of file descriptors. Note that the number of + possible connections will be slightly less than this, because + s2s itself can use some on its own. If the supply of file + descriptors is exhausted, new incoming connections will be + denied (default: 1024) -- +max_fds1024/max_fds + /io + !-- Timed checks -- check !-- Interval between checks. diff -ur jabberd-2.1.16.orig/s2s/main.c jabberd-2.1.16/s2s/main.c --- jabberd-2.1.16.orig/s2s/main.c 2007-08-31 02:24:28.0 +0200 +++ jabberd-2.1.16/s2s/main.c 2007-09-23 21:30:49.0 +0200 @@ -140,6 +140,8 @@ if (s2s-local_pemfile != NULL) log_debug(ZONE,loaded local pemfile for peer s2s connections); +s2s-io_max_fds = j_atoi(config_get_one(s2s-config, io.max_fds, 0), 1024); + s2s-check_interval = j_atoi(config_get_one(s2s-config, check.interval, 0), 60); s2s-check_queue = j_atoi(config_get_one(s2s-config, check.queue, 0), 60); s2s-check_keepalive = j_atoi(config_get_one(s2s-config, check.keepalive, 0), 0); @@ -514,7 +516,7 @@ s2s-sx_db = sx_env_plugin(s2s-sx_env, s2s_db_init); -s2s-mio = mio_new(MIO_MAXFD); +s2s-mio = mio_new(s2s-io_max_fds); s2s-retry_left = s2s-retry_init; _s2s_router_connect(s2s); Only in jabberd-2.1.16/s2s: main.c.orig diff -ur jabberd-2.1.16.orig/s2s/s2s.h jabberd-2.1.16/s2s/s2s.h --- jabberd-2.1.16.orig/s2s/s2s.h 2007-09-11 17:35:38.0 +0200 +++ jabberd-2.1.16/s2s/s2s.h 2007-09-23 21:30:49.0 +0200 @@ -106,6 +106,9 @@ /** verify-mode */ int local_verify_mode; +/** max file descriptors */ +int io_max_fds; + /** time checks */ int check_interval; int check_queue; ___ Jabberd2 mailing list Jabberd2@lists.xiaoka.com http://lists.xiaoka.com/listinfo.cgi/jabberd2-xiaoka.com
Re: [Jabberd2] jabberd2 - changes 311, 313
On Wed, Aug 01, 2007 at 09:40:31PM +0200, Christof Meerwald wrote: On Mon, 30 Jul 2007 22:14:24 +0200, Christof Meerwald wrote: just had a quick look through the recent SVN changes. I don't think the changes in mio_impl.h are correct. With the current version you will probably reference freed memory in _mio_run (which I think was the reason for the deferred closing fd in _mio_run, but I am not sure if the skip dead slots in _mio_run was needed). I have now confirmed with valgrind that 2.1.12 does reference memory that has been freed (as expected). You might, of course, also want to revert the mio_impl.h changes from change 311 (as you have at least reverted 313 now) - but then again, you could also wait another week before fixing it... Christof -- http://cmeerw.org sip:cmeerw at cmeerw.org mailto:cmeerw at cmeerw.org xmpp:cmeerw at cmeerw.org ___ Jabberd2 mailing list Jabberd2@lists.xiaoka.com http://lists.xiaoka.com/listinfo.cgi/jabberd2-xiaoka.com
