- Channel HP
- :
- Enterprise Business Blogs
- :
- Services
- :
- Technical Support Services Blog | HP Technology Services
- :
- DPTIPS: When is default security too much?
- Subscribe to RSS Feed
- Mark as New
- Mark as Read
- Bookmark
- Subscribe
- Email to a Friend
- Printer Friendly Page
- Report Inappropriate Content
DPTIPS: When is default security too much?
I should have suspected trouble was brewing when I realized how proud I was of my own cleverness: Set up a Unix IS for a customer using a Linux VM. That would eliminate the need to use one of their three HP-UX clients as an IS. The one we had anticipated using turned out to be back at 11i v1 (11.11), so it was not a viable IS platform. The remaining two 11i v3 (11.31) clients were production cluster nodes, and the powers-that-be were not receptive to the idea of tinkering with either one regardless of how benign the DP IS function might be.
No problem. One Linux VM later, we had an IS installed and patched. Now it was time to import it into the cell.
[root@neelix ~]# omnicc -import_is neelix.localdomain [12:1660] Import Installation Server failed.
When the Installation Server (IS) import failed, I said, "Aha! I know what it is. Most of the time, failed client imports are the product of inconsistent name resolution." So I quickly set about verifying that the Cell Manager (CM) could resolve the new IS by FQDN and IP address. Okay, that looks good. The problem must be with the new IS. I pop over there and verified that the new IS could resolve the CM both forwards and backwards. Well well well, now what?
Time to drop back to some basic DP troubleshooting. From the CM, can I telnet to port 5555 on the new IS? Survey says . . . . . . . . . . no, I cannot. Good. No, really -- good. This means we have something concrete to track down. Why am I unable to telnet to port 5555 on the new IS? There are no hardware firewalls in the network path between CM and IS. BUT, my default install of Linux may have included iptables. Did it? Was the service running?
[root@neelix ~]# service iptables status Table: filter Chain INPUT (policy ACCEPT) num target prot opt source destination 1 ACCEPT all -- 0.0.0.0/0 0.0.0.0/0 state RELATED,ESTABLISHED 2 ACCEPT icmp -- 0.0.0.0/0 0.0.0.0/0 3 ACCEPT all -- 0.0.0.0/0 0.0.0.0/0 4 ACCEPT tcp -- 0.0.0.0/0 0.0.0.0/0 state NEW tcp dpt:22 5 REJECT all -- 0.0.0.0/0 0.0.0.0/0 reject-with icmp-host-prohibited Chain FORWARD (policy ACCEPT) num target prot opt source destination 1 REJECT all -- 0.0.0.0/0 0.0.0.0/0 reject-with icmp-host-prohibited Chain OUTPUT (policy ACCEPT) num target prot opt source destination
Well, there's our problem! I stopped iptables momentarily and repeated my my telnet test. Now it works. So I tried my IS import again.
[root@neelix ~]# omnicc -import_is neelix.localdomain Import host successful.
There we go! I restarted the iptables service while I mulled over my options.
Does my IS really need firewall protection? Should I go to the trouble to port-limit DP and create firewall rules for the IS? In a word: NO. I'm not hosting nuclear launch codes on the IS -- it's just an install depot for Unix-flavored DP components.
The steps required to save the current iptables config, stop the service, and disable it from starting at system boot are fairly straightforward.
[root@neelix ~]# service iptables save iptables: Saving firewall rules to /etc/sysconfig/iptables:[ OK ] [root@neelix ~]# service iptables stop iptables: Flushing firewall rules: [ OK ] iptables: Setting chains to policy ACCEPT: filter [ OK ] iptables: Unloading modules: [ OK ] [root@neelix ~]# chkconfig iptables off [root@neelix ~]# chkconfig --list iptables iptables 0: off 1: off 2: off 3: off 4: off 5: off 6: off
That last command simple allowed me to verify that iptables was not set to start automatically at any runlevel.
So there you have it. Your list of likely culprits for jamming a good client import should include both faulty name resolution and firewalls. Have you overcome other roadblocks that prevented a successful client import? If so, please share your experience in the comments below!
- Mark as Read
- Mark as New
- Bookmark
- Highlight
- Email to a Friend
- Report Inappropriate Content
Alternative to iptables, host based control is used on occassions with tcp wrapper on UNIX clients, and "/etc/hosts.deny" can have entries like
"ALL : ALL"
Adding the entry
"inet : ALL"
to the "/etc/hosts.allow" file will let the client import through.
(With iptables, connection is refused upfront. With wrapper, connection is made, but service is not allowed to start.)
- Mark as Read
- Mark as New
- Bookmark
- Highlight
- Email to a Friend
- Report Inappropriate Content
Alternative to iptables, on occassions host based control is used with tcp wrapper, and /etc/hosts.deny is having entries like "ALL : ALL". Adding the entry "inet : ALL" to the /etc/hosts.allow file, would let DP communication through.
(While iptables blocks connection attempt upfront, the wrapper lets the connection made, but services, like inet in our case, are not allowed to start.)
- Mark as Read
- Mark as New
- Bookmark
- Highlight
- Email to a Friend
- Report Inappropriate Content
Hello Zoltan,
You bring up a very good case. Let me ask, would tcpwrappers without inet:ALL in /etc/hosts.allow create a condition where I could connect to 5555 on a client but not get the normal DP banner before disconnect? I ask because I had a case just this week where I installed DP 7 (omnicf and da) on an old 11.11 box. The install was error-free, and a netstat indicated that it was in fact listening on 5555. Yet when I telnet to 5555 either locally or from the network, I get a connection, a delay of several seconds, then a disconnect -- WITHOUT the normal DP banner message being displayed.
To make matters even more unusual, an attempt to import the client resulted in the CM's name being written to the client's cell_server file, but the import failed to add the client to the CM's cell_info file. Never in 14+ years have I seen behavior like this.
Any ideas?
Thanks,
Mr_T
- Mark as Read
- Mark as New
- Bookmark
- Highlight
- Email to a Friend
- Report Inappropriate Content
For the telnet 5555, and hosts.deny set, it displays exactly the behaviour you describe, i.e. connection made, then timeout without the banner. Client import would fail too, however haven't seen any tampering with cell_server file on the client side.
- Mark as Read
- Mark as New
- Bookmark
- Highlight
- Email to a Friend
- Report Inappropriate Content
Why does a link to this blog piece about Unix appear on the OpenVMS forum web page?
- Mark as Read
- Mark as New
- Bookmark
- Highlight
- Email to a Friend
- Report Inappropriate Content





