This is EAS reporting that the ISIS (Backbone) communications were unable to communicate with node 192.19.95.121 port 3675. Since EAS uses a number a range of ports of backbone commications and your cluster is "operating normally", this would indicate that when ever this communication / system update occurs something else is using this port. Since EAS will attempt to use a number of the ports in the backbone range, it will simply re-try using another port.
Try using netstat on this node to see what is using this TCP/IP port if the errors are becoming to unmanagable.
Not sure if this is a similar problem but I am getting this in the Application Log.
Backbone warning on fai-exch-slave (pid 4020) Fri Sep 28 08:20:44 2007 in ISIS_MGT_INTERCL_MODULE pr_inter.c/net_xmit(), line 912 sendto failed err=10049 destination fai-exch-master.corp.faison.com 0.0.0.0:8042
My problem is these events are followed by:
Backbone info Fri Sep 28 08:21:00 2007 : shutdown (termination of detected)
When this happens one of my nodes fails and I lose redundancy. After a few reboots (sometimes) this problem will go away and eveything will appear normal. Both nodes have experienced this problem.
This is almost always a multicast issue for me. NEVER use multicast for your domain lines. Support might tell you that it is OK and your switches/routers may support multicast but after doing literally hundreds of installs of these agents I have ALWAYS had problems if I leave the domain lines at the default of multicast. It can be a lot of work to set all the point to point lines up but you won't see your agent crash anymore.
So, select your node, go to the failure detection tab and create a point to point domain line for every NIC to every other NIC on all systems in the domain. If your domain spans a WAN then make sure that your point to point lines are using the right subnet.
For example, I had a job that had 16 nodes in the domain all HPUX. Each HPUX server had 4 NICs. Two on the public LAN, one on a backup network and another on something else. So for one node, I had a point to point domain line from each NIC to all other NICs on the other 15 servers. As you can imagine the entries were numours and it took a long time to setup but afterwards, my nodes stopped failing.
IKP1
198 Posts
0
June 22nd, 2007 09:00
This is EAS reporting that the ISIS (Backbone) communications were unable to communicate with node 192.19.95.121 port 3675. Since EAS uses a number a range of ports of backbone commications and your cluster is "operating normally", this would indicate that when ever this communication / system update occurs something else is using this port. Since EAS will attempt to use a number of the ports in the backbone range, it will simply re-try using another port.
Try using netstat on this node to see what is using this TCP/IP port if the errors are becoming to unmanagable.
IKP
scardner
1 Message
0
September 28th, 2007 08:00
Backbone warning on fai-exch-slave (pid 4020) Fri Sep 28 08:20:44 2007
in ISIS_MGT_INTERCL_MODULE pr_inter.c/net_xmit(), line 912
sendto failed err=10049 destination fai-exch-master.corp.faison.com 0.0.0.0:8042
My problem is these events are followed by:
Backbone info Fri Sep 28 08:21:00 2007
: shutdown (termination of detected)
When this happens one of my nodes fails and I lose redundancy. After a few reboots (sometimes) this problem will go away and eveything will appear normal. Both nodes have experienced this problem.
Any ideas?
Win2k3 Servers SP1
Exchange 2003 SP2
CSBSAA 5.1.2
Ryan9
45 Posts
0
September 28th, 2007 08:00
So, select your node, go to the failure detection tab and create a point to point domain line for every NIC to every other NIC on all systems in the domain. If your domain spans a WAN then make sure that your point to point lines are using the right subnet.
For example, I had a job that had 16 nodes in the domain all HPUX. Each HPUX server had 4 NICs. Two on the public LAN, one on a backup network and another on something else. So for one node, I had a point to point domain line from each NIC to all other NICs on the other 15 servers. As you can imagine the entries were numours and it took a long time to setup but afterwards, my nodes stopped failing.