Hi Everyone, Problem has been SOLVED. My colleague was playing with hearbeat on this machine and didn't forget to reset net.ipv4.ip_nonlocal_bind = 0. I fixed it, did sysctl -p and now: # gluster peer probe 192.168.1.193 Probe successful Thanks, guys, for great support and sorry for keep you busy ;-) BR, Vladimir. On Mon, Jul 2, 2012 at 6:34 PM, Vladimir Yakovlev <mrquesty at gmail.com>wrote: > Hi Philip, > > I tried IPs instead of hostnames at the beginning, but it didn't help: > > [root at host1 ~]# gluster peer probe 192.168.1.193 > Probe on localhost not needed > > My assumption is smth wrong smwhere else. Donno where - no ideas at all; > the worst is i have a set of test machines, where everything works > perfectly... :( > > There are two switches between these machines, but they allow Gluster > traffic responsively. > > Hope, we can struggle this together ;-) > > Thanks, guys, for quick responses. > > Regards, > vy > > On Mon, Jul 2, 2012 at 6:27 PM, Philip Poten <philip.poten at gmail.com>wrote: > >> Hi Vladimir >> >> To circumvent any remaining /etc/hosts abnormalities, try with IPs >> only. if that doesn't work either, you have a problem outside name >> resolution. If it works with IPs, the problem is the resolver. >> >> Note that nscd caches answers, also the "host" command uses DNS >> responses, /not/ the /etc/hosts (nsswitch.conf order resp.) file - but >> gluster does, as does "ping" and anything else using gethostbyname(3) >> >> hth, >> Philip >> >> 2012/7/2 Vladimir Yakovlev <mrquesty at gmail.com>: >> > Thanks, Brian, for the hint. >> > I've changed /etc/hosts with respect to your comment, but it didn't help >> > either. >> > >> > The problem (from my perspective) in smth else, e.g. why, when I try to >> do >> > the following, I see the blank response in tcpdump: >> > [root at host1 ~]# ip a >> > 1: lo: <LOOPBACK,UP,LOWER_UP> mtu 16436 qdisc noqueue state UNKNOWN >> > link/loopback 00:00:00:00:00:00 brd 00:00:00:00:00:00 >> > inet 127.0.0.1/8 scope host lo >> > 2: eth0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc pfifo_fast >> state >> > DOWN qlen 1000 >> > link/ether 00:25:90:30:34:42 brd ff:ff:ff:ff:ff:ff >> > inet 46.../27 brd 46.182.25.159 scope global eth0 >> > 3: eth1: <BROADCAST,MULTICAST,UP,LOWER_UP> mtu 1500 qdisc pfifo_fast >> state >> > UP qlen 1000 >> > link/ether 00:25:90:30:34:43 brd ff:ff:ff:ff:ff:ff >> > inet 192.168.1.192/24 brd 192.168.1.255 scope global eth1 >> > [root at host1 ~]# tcpdump -i eth1 'port 27004' >> > tcpdump: verbose output suppressed, use -v or -vv for full protocol >> decode >> > listening on eth1, link-type EN10MB (Ethernet), capture size 65535 bytes >> > ^Z >> > [2]+ Stopped tcpdump -i eth1 'port 27004' >> > [root at host1 ~]# bg >> > [2]+ tcpdump -i eth1 'port 27004' & >> > >> > [root at host1 ~]# gluster peer probe host2 >> > Probe on localhost not needed >> > [root at host1 ~]# fg >> > tcpdump -i eth1 'port 27004' >> > ^C >> > 0 packets captured >> > 117 packets received by filter >> > 52 packets dropped by kernel >> > >> > So (by whatever reason) Gluster doesn't send a packet to anywhere >> through >> > ethernet. >> > >> > Guys, any ideas? >> > >> > Thanks, >> > BR, >> > vy >> > >> > >> > On Mon, Jul 2, 2012 at 6:06 PM, Brian Candler <B.Candler at pobox.com> >> wrote: >> >> >> >> On Mon, Jul 02, 2012 at 05:54:46PM +0400, Vladimir Yakovlev wrote: >> >> > I tried different configurations, the latest follows: >> >> > [root at host1 ~]# cat /etc/hosts >> >> > 127.0.0.1 localhost localhost.localdomain localhost4 >> >> > localhost4.localdomain4 >> >> > 127.0.0.1 host1 host1.fqdn >> >> > 192.168.1.193 host2 host2.fqdn >> >> > [root at host2 ~]# cat /etc/hosts >> >> > 127.0.0.1 localhost localhost.localdomain localhost4 >> >> > localhost4.localdomain4 >> >> > 127.0.0.1 host2 host2.fqdn >> >> > 192.168.1.192 host1 host1.fqdn >> >> >> >> Be consistent. Both machines should have: >> >> >> >> 127.0.0.1 localhost localhost.localdomain >> >> 192.168.1.192 host1.fqdn host1 >> >> 192.168.1.193 host2.fqdn host2 >> >> >> >> On host1 you should have "host1.fqdn" in /etc/hostname, and the command >> >> "hostname" should show it. (Ditto for host2, but with "host2.fqdn" of >> >> course) >> > >> > >> > >> > _______________________________________________ >> > Gluster-users mailing list >> > Gluster-users at gluster.org >> > http://gluster.org/cgi-bin/mailman/listinfo/gluster-users >> > >> > > -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://gluster.org/pipermail/gluster-users/attachments/20120702/468354fa/attachment.htm>
