qstat: cannot connect to server ... (errno=111)

I could pass this error only when commented lines

pbs 15000/tcp
pbs_dis 15001/tcp
pbs_dis 15001/udp
pbs_mom 15002/tcp
pbs_mom 15003/udp
pbs_mom 15003/tcp
pbs_sched 15004/tcp

in ng2:auckland.ac.nz:/etc/services file.
I placed those line reading Vladimir's recommendations.

UPD: after some tests I've found out that it's enough to edit first line in this group. It should be pbs_server 15000/tcp and then pbs utilities can connect to pbs_server. Even if on the box where pbs_server is running this line should like that, otherwise pbs utilities don't see the pbs_server.

unknown nfs status return value: -1

This error message might appear while mounting NFS share exported from AIX box.
That means an importer doesn't have a record in DNS database.

To solve just add an entry about the importer into /etc/hosts file.

Source...

Virtual File System Manager

Some things to remember while using Virtual File System Manager:

1. It looks like in Windows environment VFSM requires domain name of remote host should be A record rather than CNAME; If domain name of a host is CNAME, VFSM can't connect to it with an error like this:
2. In Linux environment VFSM can mount remote file system of a host with both kind of domain names (A record or CNAME);
3. VFSM can copy files between remote file system and local file system and back and can't copy files between two remote file systems. If user copies files between two remote hosts then files aren't appeared on destination hosts but deleted from source host.
   
4. To copy user needs to use shortcuts Ctrl+C and Ctrl+V instead of mouse drag-n-drop.
5. If there was an attempt to get access to any folder but /tmp as Anonymous user then VFSM didn't give access even to /tmp folder. The best way I know is to restart VFSM.
6. May be there are some more troubles but I didn't investigate them yet.
   

CRL has expired

That means Certificate Revocation List has been expired. This error may appear when create proxy with -verify option:

grid-proxy-init -verify -debug

It's a file in /etc/grid-sequrity/certificates/1e12d831.r0 (for APACGrid CA) Next Update is overdated. Current (for the date of this posting) CRL has inside:

Version 1 (0x0)
Signature Algorithm: md5WithRSAEncryption
Issuer: /C=AU/O=APACGrid/OU=CA/CN=APACGrid/emailAddress=camanager@vpac.org
Last Update: Jun 18 23:50:25 2007 GMT
Next Update: Jul 18 23:50:25 2007 GMT

It hasn't been updated on my bestgrid-02 machine automatically because on unknown reason a line to call fetch-crl.cron script disappeared from crontab file.

To update APACGrid CA CRL use link http://ca.apac.edu.au/pub/crl/cacrl.txt
and place this file as /etc/grid-sequrity/certificates/1e12d831.r0. Also to automatic updates it's a good idea to create cron task for daily or even hourly updates of CRL:

#!/bin/bash

# /opt/vdt/fetch-crl/share/doc/fetch-crl-2.6.2/fetch-crl.cron
# in crontab:
# 12 2 * * * /opt/vdt/fetch-crl/share/doc/fetch-crl-2.6.2/fetch-crl.cron
# Designed to be ran once a day.

if [ -f /etc/sysconfig/fetch-crl ] ; then
. /etc/sysconfig/fetch-crl
fi

CRLDIR=${CRLDIR:-/opt/vdt/globus/TRUSTED_CA/}

QUIET=${QUIET:-yes}

if [ $QUIET = "yes" ] ; then
/opt/vdt/fetch-crl/sbin/fetch-crl --loc $CRLDIR -out $CRLDIR -a 24 --quiet
else
/opt/vdt/fetch-crl/sbin/fetch-crl --loc $CRLDIR -out $CRLDIR
fi

Submitting Job to Run in Canterbury

To submit job on ng2.canterbury.ac.nz I needed to run ggc (https://www.seegrid.csiro.au/twiki/bin/view/Compsrvices/GenericGridGUI) and fill fields like that.









It's possible to see a new folder with output files using Virtual File System Manager (http://www.vpac.org/twiki/bin/view/Sandbox/UsingVFSManager).

More detailed description is in http://www.bestgrid.org/index.php/Getting_Started_with_BeSTGRIDs_Computational_GRID

Transport e-mails to Sakai

Sakai has own SMTP server - James. To receive messages in Sakai mailhost.auckland.ac.nz should forward messages to sakai.bestgrid.org. Now MailHost forwards messageg to services.bestgrid.org. It's a unneeded chain ring.
In sakai.properties set parameters for e-mails and use any dns IPs.

ToDo - ask Service Desk to redirect messages for *.bestgrid.org to sakai machine.

Adding user to Samba on AIX

edit file
/opt/pware/samba/3.0.24/lib/smb.conf

for each user add something like that
# A private directory, usable only by fred. Note that fred requires write
# access to the directory.
;[fredsdir]
; comment = Fred's Service
; path = /usr/somewhere/private
; valid users = fred
; public = no
; writable = yes
; printable = no

Create user account and set password:
/opt/pware/samba/3.0.24/bin/smbpasswd -a userupi

Job submission (globusrun-ws)

- copy usercert.pem into ~/.globus folder
- copy userkey.pem into ~/.globus folder and change permissions to 600
- run grid-proxy-init -debug
- run globusrun-ws -Ft PBS -submit -S -c /bin/date

ng2 machine should be in acl_hosts list of PBS server (qmgr -c 'l s') and in /etc/hosts.equiv file

Xen error: Link veth4 is missing

On default Xen has four veths to communicate between bridges and real eths. If one uses a command:

/etc/xen/scripts/network-bridge start netdev=eth4 bridge=xenbr4 vifnum=4

to create bridge xenbr4 and connect it to eth4 then he receives a message with
Link veth4 is missing at the beginning. To fix that situation gateway.auckland.ac.nz:/boot/grup/grub.conf should be changed like this:

kernel /boot/xen-3.gz dom0_mem=262144
module /boot/vmlinuz-2.6.16.33-xen ro root=LABEL=/ netloop.nloopbacks=8
module /boot/initrd.img-2.6.16.33-xen

A number in netloop.nloopbacks parameter is equal to amount of real NICs. Real gateway server shoud be rebooted.

Automate VM creation

To automate creation of VMs I developed a script based on VPAC recomendations:

#!/bin/sh

#Create a VM:
# allocate logical volumes for root and swap filesystems
# make swap and root filesystems
# untar VM image
# call editor to edit network configuration files
# copy and edit xen template
# create VM
#
#Parameters:
# $1 - Size for root filesystem
# $2 - Name of VM
#
#Author: Andrey Kharuk, 01/06/2007
#

if [ -z "$1" ]; then
echo " " >&2
echo "Usage: `basename $0` disksize(in Mb) VMname" >&2; exit 2
echo " " >&2
echo " " >&2
fi

lvcreate -L $1M -n $2R VolumeGroup00 ;
lvcreate -L 512M -n $2S VolumeGroup00 ;
mkswap /dev/VolumeGroup00/$2S ;
mkfs -t ext3 /dev/VolumeGroup00/$2R ;
mkdir /srv/$2 ;
mount -t ext3 /dev/VolumeGroup00/$2R /srv/$2 ;
cd /srv/$2 ;
tar xjpf /etc/xen/vm-image-0.0.1.tar.bz2 ;
nano /srv/$2/etc/sysconfig/network ;
nano /srv/$2/etc/sysconfig/network-scripts/ifcfg-eth0 ;
cp /etc/xen/template /etc/xen/$2 ;
nano /etc/xen/$2 ;
cd ..
umount /srv/$2 ;

xm create -c $2

Tar-ball 'vm-image-0.0.1.tar.bz2' is an image of fresh installed CentOS 4.0 on VM.
Both files reside in gateway.bestgrid.org:/etc/xen