[GE users] FW: [GE users] shadow host configuration

thamizhannal Thamizhannal.Paramasiuam at Honeywell.com
Thu May 27 10:15:22 BST 2010

    [ The following text is in the "Windows-1252" character set. ]
    [ Your display is set for the "ISO-8859-10" character set.  ]
    [ Some special characters may be displayed incorrectly. ]


We have stopped the sge_shadowd service on the master host. The shadow service is running only on the shadow node.

We couldn?t find any error on the /tmp folder.

Below are the contents of scheduler log message.

05/25/2010 14:47:59|schedd|mgr|I|starting up SGE 6.1u2 (lx24-amd64)
05/25/2010 14:48:59|schedd|mgr|E|commlib error: can't connect to service (Connection refused)
05/25/2010 14:50:00|schedd|mgr|I|controlled shutdown 6.1u2
05/25/2010 14:50:15|schedd|mgr|I|starting up SGE 6.1u2 (lx24-amd64)

For testing, we have killed the qmaster service running on the manager node, but the shadow host doesn?t take the control of the manager.

The log shows the manager service is ?controlled shutdown?.  Is there any other way to test the shadow host?

Since the machines are in remote location we couldn?t able to poweoff or remove the network to test it.

Kindly help us in configuring and testing the shadow host.


From: thamizhannal [mailto:Thamizhannal.Paramasiuam at Honeywell.com]
Sent: Monday, May 24, 2010 2:58 PM
To: users at gridengine.sunsource.net
Subject: [GE users] shadow host configuration


We have configured a shadow host for our existing SGE cluster.

But when the qmaster service is killed, the shadow host doesn?t take up the qmaster?s job.

It doesn?t show any error in the spool directory.

The following are the list of services running on master and shadow host.

Master host:

root       426     1 0 May20 ?        00:00:00 /opt/sge/bin/lx24-amd64/sge_shadowd
root     31636     1  0 13:16 ?        00:00:00 /opt/sge/bin/lx24-amd64/sge_qmaster
root     31655     1  0 13:16 ?        00:00:00 /opt/sge/bin/lx24-amd64/sge_schedd

Shadow host:
root     21632     1 0 04:44 ?        00:00:00 /opt/sge/bin/lx24-amd64/sge_shadowd

Both the master and shadow host are NFS mounted.

Whether the shadow host also needs to run the qmaster and schedd service? Please help me in configuring shadow host.

As per my understanding, only shadowd service should run on the shadow host.

Kindly let me know whether my understanding is correct.


More information about the gridengine-users mailing list