Thoughts on Azure, OMS & SCOM: OM12 SP1 UNIX/Linux Agent Troubleshooting Table

Monday, June 17, 2013

OM12 SP1 UNIX/Linux Agent Troubleshooting Table

Even though there are already many good postings and articles out there, all about this topic (will list them at the end of this posting) I still want to add my experiences as well.

For a customer quite a few Linux servers had to be monitored. During the roll-out of these OM12 SP1 Agents to the Linux systems several errors popped up. Thanks to a highly experienced Linux guru working for this customers these issues were sorted out pretty fast. Based on this experience I have made a table with the most occurring errors and their possible causes and their fixes.

Issue	Cause & Resolution
DNS Configuration error	01: Faulty reverse DNS Lookup Zone. When fixed all went just fine 02: Linux system had multiple names, all registered in DNS. After a couple of retries the Agent landed properly. 03: System resided in an old segment which didn’t have a zone on the new DNS servers. When fixed all went just fine
Failed during SSH Discovery	01: SSH was locked down to ROOT only. When fixed for the OM12 SP1 account used by Linux all went just fine. 02: An outdated version of SSH which isn’t compatible with the .NET SSH implementation Microsoft uses on the OM12 SP1 side. SSH requires an update. 03: An outdated version of SSH which doesn’t accept certain SSH calls. SSH requires an update.
Failed to install kit	01: Home folder of the OM12 SP1 Linux account was missing. After having added this folder all went just fine. 02: Certain files were locked. When retried the installation of the OM12 SP1 Agent some hours later all went just fine.
Installation hangs	On some systems the installation of the OM12 SP1 Linux Agent just hanged. Had to hard stop the OM12 SP1 Console. Then a second attempt went just fine.
Unexpected Discovery Result	01: Reason unknown. Second attempt (some hours later) ran just fine. 02: A restart of the OM12 SP1 services on the OM12 SP1 MS running the Discovery (be careful though): http://www.opsman.co.za/?p=50
WinRM cannot complete the operation	Firewall was blocking WinRM service. After having opened that port (TCP 1270) it still didn’t work. See this posting to get it working: http://blogs.technet.com/b/chandanbharti/archive/2011/12/21/linux-agent-install-issue.aspx
Agent verification failed	Multiple DNS issues: 1: Linux system has a different hostname compared to the FQDN. Correct it (hostname or FQDN) and all is just fine. 2: DNS record isn’t present. Add the record and all is just fine.