Tue Oct 19, 2021 12:15 am
Login Register Lost Password? Contact Us


Multi node installation problem

Post questions specific to installation or configuration for the HPCC Systems platform

Fri Aug 19, 2011 6:37 am Change Time Zone

Hello,

I want to evaluate HPCC. I have 3 physical servers with 24core cpu, 256g memory each.

First I try one node installation which work fine. Then I installed the system to the other nodes and use configuration wizard to make them work together. I restart all hpcc services each node.

However when I try to login to ECL IDE it gives me SOAP error message. I even couldn't login to EclWatch web page.

I guess the problem is our network interfaces. The servers are connected via infiniband in private(bond1), and ethernet in public network(bond0).

ex:
Below bond0 has public ip and bond1 has private ip. We are not using eth0 at all.
In the environment.xml I've changed the interface parameter eth0 to bond0 but it didn't work. Any idea?

bond0 Link encap:Ethernet HWaddr 18:A9:05:59:FE:80
inet addr:10.210.232.91 Bcast:10.210.232.255 Mask:255.255.255.0
inet6 addr: fe80::1aa9:5ff:fe59:fe80/64 Scope:Link
UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
RX packets:10009342 errors:0 dropped:0 overruns:0 frame:0
TX packets:817714 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:5830590692 (5.4 GiB) TX bytes:94781126 (90.3 MiB)

bond1 Link encap:InfiniBand HWaddr 80:00:00:48:FE:80:00:00:00:00:00:00:00:00:00:00:00:00:00:00
inet addr:182.2.1.3 Bcast:182.2.255.255 Mask:255.255.0.0
inet6 addr: fe80::223:7dff:ff94:5201/64 Scope:Link
UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1
RX packets:132893 errors:0 dropped:0 overruns:0 frame:0
TX packets:117473 errors:0 dropped:1 overruns:0 carrier:0
collisions:0 txqueuelen:0
RX bytes:7183908 (6.8 MiB) TX bytes:5317338 (5.0 MiB)
onur
 
Posts: 2
Joined: Fri Aug 19, 2011 6:18 am

Mon Aug 22, 2011 3:53 pm Change Time Zone

Couple of things you can try:

1. Make sure the hostname of the machine(s) in question contain an ip listing for the ip you have listed for that machine in your config.

2. On our IB systems our infiniband interface is listed as ib0 (or ib1) and it has no issue bonding to that interface as long as #1 above is done. Same will work for bonding, just make sure your hostname entry in /etc/hosts lists the ip of the bonding interface.

3. Validate that it is indeed a binding issue by running a netstat -tanp and validating service is not bound to a different adapter.

4. See if the services are running at all in your process list to make sure you don't have any other type of startup issues.
jonburger
 
Posts: 8
Joined: Thu Jun 16, 2011 6:21 pm


Return to Installation

Who is online

Users browsing this forum: No registered users and 1 guest

cron