This post is more than 5 years old
2 Intern
•
222 Posts
0
8702
naviagent won't start on RHEL 5.7 host
I have several RHEL hosts, some run the naviagent fine, others won't start for unknown reasons. I could use some advice on how to get naviagent running on the boxes which currently won't run the agent. Some of these hosts run Powerpath, some run native Linux multipathing. Details below.
[root@hp1 ~]# cat /etc/redhat-release
Red Hat Enterprise Linux Server release 5.7 (Tikanga)
[root@hp1 ~]# uname -srvmo
Linux 2.6.18-274.12.1.el5 #1 SMP Tue Nov 8 21:37:35 EST 2011 x86_64 GNU/Linux
[root@hp1 ~]# /etc/init.d/naviagent status
naviagent is stopped
[root@hp1 ~]# /etc/init.d/naviagent start
Starting Navisphere agent: [FAILED]
[root@hp1 ~]# /opt/Navisphere/bin/naviagent -f /etc/Navisphere/agent.config
[root@hp1 ~]# echo $?
1
[root@hp1 ~]# grep -i navi /var/log/messages
Jul 18 04:20:13 hp1 naviagent[19447]: Agent Main -- Net or File event. Err: EMULSocket::bind()
christopher_ime
2K Posts
0
July 27th, 2012 17:00
Seeing the information you pasted, you have both the more recent Unisphere Host Agent and the older Navisphere Agent installed on the Linux host. Unlike Windows, Linux will not prevent you from doing so (or in other words, Windows will detect the transition from Navisphere Agent to Unisphere Hostagent as an upgrade). There is no reason to have both installed on the server.
Since you already have hostagent running and bound to port TCP/6389, your attempts to start naviagent will fail as noted by the bind() error message in the kernel logs. I would uninstall naviagent.
Please note, there isn't any reason you have to match the version of the agent with the version of Navisphere Manager or Unisphere Manager running on the storage processors. For instance, the latest version of Unisphere Host Agent v1.2 provides the following details:
/etc/Unisphere/SupportedFlareRevisions
[...]
CX_Software >= 2.26 AND CX_Software < 2.33
[...]
Also, if you visit support.emc.com and select either a CX3 or CX4 as your array model and select Downloads, you will see the latest Unisphere Host Agent in the list (assume that the site is properly filtering only the relevant choices based on your array, but worth double-checking).
Lastly, not related to the original question, your agent.config file (/etc/Unisphere) is missing entries for privileged users (local or remote) that can run CLI commands against the agent. Notably you are missing reference to both of the storage processors which query the agent on a regular interval. Add the following to the file:
user system@
user system@
<blank line or in other words CR on last line if EOF>
As a reminder, any changes to configuration files will require restarting the agent. This may explain why you are gettings certain details from one but not the other.
HTH
usao
2 Intern
2 Intern
•
222 Posts
0
July 18th, 2012 06:00
The hosts which in which the agent will not start are a mix of native linux multipath daemon and powerpath. There is a "default" agent.config file, which I have not modified, on every host.
The problem is not specific to which multipath daemon im using.
I have hosts with working agents using both multipath and powerpath, and hosts which are not able to run the agent using both multipath and powerpath.
It so happens that the host I showed in my initial post is using powerpath.
The output of the default agent.config is posted below, with the comments removed to make the output shorter.
[root@hp1 ~]# sed 's/#.*$//' /etc/Navisphere/agent.config | grep -v '^$'
clarDescr Navisphere Agent
clarContact John Smith, 800-555-1212
device auto auto
user root
poll 60
eventlog 100
baud 9600
cris_danci
39 Posts
0
July 18th, 2012 06:00
Is the host multihomed? If so you need to created an agentID.txt file.
Also please provide the output of:
cat /etc/Navisphere/agent.config
usao
2 Intern
2 Intern
•
222 Posts
0
July 18th, 2012 07:00
Im not aware of what the procedure generator is. Can you point me to more details regarding this?
cris_danci
39 Posts
0
July 18th, 2012 07:00
Multipating has nothing to do with it. The agent has no awareness of the multipath driver in operations.
The agent file looks ok (other than the fact it won't work because you don't have user entries). Is the host multihomed? Does it have more than one external network interface?
cris_danci
39 Posts
0
July 18th, 2012 07:00
You;ll need an agentID.txt file. Check the procedure generator
usao
2 Intern
2 Intern
•
222 Posts
0
July 18th, 2012 07:00
Each host has 2 GBE cards. As far as the network ports in-use, we have 2 seperate networks, one internal and one DMZ. This is true for both hosts which have working and non-working agents.
Example of host with working agent:
[root@hp2 ~]# ifconfig | egrep 'Link|inet'
eth0 Link encap:Ethernet HWaddr 00:1E:0B:CE:58:50
inet addr:10.0.9.20 Bcast:10.0.255.255 Mask:255.255.0.0
inet6 addr: fe80::21e:bff:fece:5850/64 Scope:Link
eth1 Link encap:Ethernet HWaddr 00:1E:0B:CE:58:52
inet addr:10.0.0.20 Bcast:10.0.255.255 Mask:255.255.0.0
inet6 addr: fe80::21e:bff:fece:5852/64 Scope:Link
Example of host with non-working agent:
usao
2 Intern
2 Intern
•
222 Posts
0
July 18th, 2012 07:00
I manually created an agentID.txt file based on notes I found on google as follows, but it still fails to start. I checked the host with working naviagent, and it also did not have an agentID.txt file.
[root@hp1 ~]# cat /agentID.txt
hp1.dssolutions.com
10.0.9.10
[root@hp1 ~]# /etc/init.d/naviagent start
Starting Navisphere agent: [FAILED]
[root@hp1 ~]#
usao
2 Intern
2 Intern
•
222 Posts
0
July 18th, 2012 07:00
Also, I am including the output of the HostIdFile.txt on both working and non working agent hosts.
Working agent host:
[root@hp2 ~]# cat /var/log/HostIdFile.txt
10.0.9.2020110408123916
Nonworking agent host:
[root@hp1 ~]# cat /var/log/HostIdFile.txt
10.0.9.1020111206023525
cris_danci
39 Posts
0
July 18th, 2012 16:00
1. Check that the ports the agent is trying to use isn't already in use using netstat
2. Disable any firewalls and try again.
3. Generate the document using the online generator https://mydocs.emc.com/VNX/requestMyDoc.jsp
4. Register for a powerlink account (http://powerlink.emc.com) and check primus
christopher_ime
2K Posts
0
July 18th, 2012 19:00
Oops, sorry cris, I see that you already asked. Ignore me.
christopher_ime
2K Posts
0
July 18th, 2012 19:00
usao,
Using netstat, can you check to see if TCP/6389 might already be bound by another service/process? That could also be why it is failing to start.
usao
2 Intern
2 Intern
•
222 Posts
0
July 26th, 2012 07:00
I did find the following entry in the /var/log/messages on the host which the agent fails:
Jul 26 07:49:43 hp1 naviagent[26187]: Agent Main -- Net or File event. Err: EMULSocket::bind()
Again, both servers are right next to each other, on the same network, no firewalls between the hosts or between the hosts and the array.
usao
2 Intern
2 Intern
•
222 Posts
0
July 26th, 2012 07:00
After a bit of a delay working other issues, im back to this problem.
I have generated the /agentID.txt files on both hosts. Then I removed the /var/log/HostIdFile.txt and tried to restart the naviagent. Again, one host works the other fails. Neither host is making use of port 6389.
[root@hp2 ~]# netstat | grep 6389
[root@hp2 ~]# /etc/init.d/naviagent restart
Shutting down Naviagent: [ OK ]
Starting Navisphere agent: [ OK ]
[root@hp2 ~]#
[root@hp1 ~]# netstat | grep 6389
[root@hp1 ~]# /etc/init.d/naviagent restart
Shutting down Naviagent: [FAILED]
Starting Navisphere agent: [FAILED]
[root@hp1 ~]#
Is there a log somewhere which would indicate why it's failing?
dynamox
2 Intern
2 Intern
•
20.4K Posts
0
July 26th, 2012 07:00
Any services that are running and using the same port as navi agent, as pointed out by others ?