Redundant servers in FLEXlm

Accelerated Computing HPC Compilers Legacy PGI Compilers
1

Hi I’m trying to get 3 servers to offer our PGI license in a redundant quorum. However, it seems that all 3 servers MUST be up for the license service to work correctly.

It seems as though if the license server is down on ANY other redundant server, the license server cannot be started.

My 3 servers are a Red Hat box, Windows Server 2003 box, and a Sun Solaris box.

For the purposes of this post, consider:

Red Hat = ServerA
Windows = ServerB
Solaris = ServerC

With ALL 3 machines NOT running FlexLM, I start lmgrd.rc on ServerA and get:

11:20:47 (lmgrd) FLEXnet Licensing (v11.4.100.0 build 50818 i86_se9) started on ServerA (linux) (2/25/2008)
11:20:47 (lmgrd) Copyright (c) 1988-2007 Macrovision Europe Ltd. and/or Macrovision Corporation. All Rights Reserved.
11:20:47 (lmgrd) US Patents 5,390,297 and 5,671,412.
11:20:47 (lmgrd) World Wide Web: http://www.macrovision.com
11:20:47 (lmgrd) License file(s): /opt/pgi/linux86/7.1-5/bin/license.dat
11:20:47 (lmgrd) lmgrd tcp-port 27000
11:20:47 (lmgrd) Redundant server configuration enabled. Attempting to establish quorum…
11:20:47 (lmgrd) Trying to establish connections to other redundant servers…
11:20:47 (lmgrd) Attempting connection to secondary redundant server ServerB
11:20:47 (lmgrd) Attempting connection to tertiary redundant server ServerC
11:20:52 (lmgrd) Connection attempt to secondary redundant server ServerB failed
11:20:52 (lmgrd) Connection attempt to tertiary redundant server ServerC failed
11:21:17 (lmgrd) Attempting connection to secondary redundant server ServerB
11:21:17 (lmgrd) Attempting connection to tertiary redundant server ServerC
11:21:22 (lmgrd) Connection attempt to secondary redundant server ServerB failed
11:21:22 (lmgrd) Connection attempt to tertiary redundant server ServerC failed
…and so on.

If I check lmutil lmstat on ServerA, I see:

[root@ServerA bin]# lmutil lmstat
lmutil - Copyright (c) 1989-2007 Macrovision Europe Ltd. and/or Macrovision Corporation. All Rights Reserved.
Flexible License Manager status on Mon 2/25/2008 11:24

License server status: 27000@ServerA,27000@ServerB,27000@ServerC
License file(s) on ServerA: /opt/pgi/linux86/7.1-5/bin/license.dat:

ServerA: license server UP (MASTER) v11.4
ServerB: Cannot connect to license server system. (-15,570:115 “Operation now in progress”)
ServerC: Cannot connect to license server system. (-15,570:115 “Operation now in progress”)

Vendor daemon status (on ServerA):

pgroupd: Cannot connect to license server system. (-15,570:115 “Operation now in progress”)

[root@ServerA bin]#


My question: WHY will pgroupd NOT start? It seems like pgroup will not start because other redundant servers are not running. If so, what’s the point of having a redundant config if the other servers MUST run?

Thanks,
Phil

2

Phil,

Sorry you are having problems. Redundant servers should work if 2 of the 3
are up. I think all three must be present when you start up lmgrd. We have never
tried your configuration, and do not intend to.

We do not recommend using them because servers are much more reliable
nowadays, and powerful enough to keep up with the license requests.

We cannot diagnose site problems with them, and we mainly provide the
ability to create them because sites like Boeing have always used them and
continue to use them, and have the IT expertise needed to support them.

Make sure firewalls are not preventing the ports used by lmgrd and pgroupd
are not being blocked, and that the servers can communicate. Also make
sure the ‘user’ running lmgrd and pgroupd has proper permissions
in the directories they are running from.

I assume you got the proper daemons for each of the systems. They can be
found at https://www.pgroup.com/support/downloads.php.

Many users have changed their approach, and created a second single
server license to use in the case their main server fails. This is okay as long
as the two servers are not working at the same time, and the second
license is for emergency purposes.