[GPFS] Adding a disk to the filesystem concurrently

[GPFS] Adding a disk to the GPFS filesystem concurrently It's very simple to add a disk to the filsystem.
We will make it from making nsd to add a disk to the filesystem.

1. Check the disk status on both nodes (check disk number, pvid)
# lspv

2. For adding a 'nsd' to GPFS cluster, making a disk.desc file when we use it before the GPFS configuration

# vi diskadd_desc
hdisk6:::dataAndMetadata:-1:kn_nsd3

3. Adding a nsd using a 'mmcrnsd' command mmcrnsd.

# mmcrnsd -F /diskadd_desc


Check whether the nsd is well made or not

# mmlsnsd



4. Adding the nsd we made to the filesystem.

# mmadddisk filesystem_name nsd_name

Now you can see the changed filesystem using 'mmlsnsd' and 'mmdf' command.

# mmlsnsd
# mmdf filesystem_name


5. Finally, this job will be completed after using 'restripe' command for redistribute data.
  (주의!!! This command can use in operating status but I strongly recommend not to use it at busy operating. It can make a performance issue.)
# mmrestripefs filesystem_name -b

After upper action, you can check the filesystem using below command.
# mmdf filesystem_name

[GPFS] Authenticate ssh, rsh before GPFS configuration

Crontab configuration.

 It is easy and powerful utility of schuduling jobs.
 And crontab is using variable OS like linux, solaris and so on.

1. Contents of crontab
File Location: /var/spool/cron/crontabs/         # Restoring each file for each account
Log Location: /var/adm/cron/log

Listing current crontab configuration

# crontab -l

2. A command for modifying crontab
# crontab -e                # It uses same editor as vi.

3. Control using crontab command

       /var/adm/cron/cron.allow : User lists allow to use it

       /var/adm/cron/cron.deny : User lists not to allow to use it

       If there isn't 'cron.allow, cron.deny'  file, only root user can use it.

4. Form of crontab contents

Minute     hour     day_of_month     month     weekday     command

ex) 30 6 * * 1,3,5 /usr/bin/calendar        # Excute calendar at every 06:30am mon, tue, fri every month

Minute: 30

hour: 6

day of month: *(every day)

month: *(every month)

weekday: 1,3,5 (Sunday through Saturday), mon, tue, fri

4. Kill the crontab daemon

Check crontab daemon 

# ps -ef|grep cron

Type the below command and check again.

 
# ps -ef |grep cron | grep -v grep | awk '{print "kill -9 "$2}' | sh -x

[AIX] How to make 'alt disk' using alt_disk_install

Let me introduce to you about making alt disk for system backup and test server system or new server instead of mksysb.

1. Prepare available disk(hdisk2)

root@test: / # lspv
hdisk0          00ce30ffea377c7b                    rootvg          active
hdisk1          00ce30ff5eac4b0a                    rootvg          active
hdisk2          none                                       none
 
1.1 Making pvid using chpv command

root@test: / # chdev -l hdisk2 -a pv=yes
hdisk2 changed

root@test: / # lspv
hdisk0          00ce30ffea377c7b                    rootvg          active
hdisk1          00ce30ff5eac4b0a                    rootvg          active
hdisk2          00ce30fffea34bc7c                    none

2. 'alt disk' is reference of 'image.data' file smilar with mksysb backup.
So you should modify 'image.data' file when you use it to mirrored rootvg.
It just can use only changing LP, PP and copy contents to 1 of image.data file and using below command.

root@test: / # alt_disk_install -CBO hdisk2 -i image.data_name

The option of alt_disk_install we usually used are listed below.
 
- C : clone
 
- B : Not to change bootlist
  
- O : altinst_rootvg ODM cleaning for using another system

       Cleaning ODM contents  with option O
      - hostname, N/W information,  other device information




Inform)

When you try to use 'alt disk' for other system, you should to input option 'O'.
But remove the option when you use it for same system for system backup.

3. You can see an alt disk after complete upper task.

root@test: / # lspv
hdisk0          00ce30ffea377c7b                    rootvg          active
hdisk1          00ce30ff5eac4b0a                    rootvg          active
hdisk2          00ce30fffea34bc7c                   altint_rootvg

4.Remove the alt disk from system. Type the below command. (Not to use exportvg command)

#alt_disk_install -X altinst_rootvg

If you don't use upper command when moving another system, The system's ODM could be complicated

Sometimes you need to change 'bootlist' because it can be changed to altinst_rootvg

5. You can use a below command not mirrored vg but extended vg.

#smitty alt_clone을 이용하거나
 
Or changing image.data likes number 2 and type below command
 
#alt_disk_install -CBO -i /image.data hdisk2

[aix] time setting on aix using xntp daemon

How to make a time sync. between systems

We need to make a time sync. between clustering system and many database(DB) server.

There are big 3-ways that we can make a time sync.

1. This is very simple command using 'setclock'

It makes a time sync only once with host so can be different as time goes by.

you can use 'crontab', schedule daemon, if you want to set up time every time you want to.

I'll introduce to you how to use crontab daemon later

2. Using timed daemon

- Synchronize servers' time on the same network(LAN).

- It can make huge gaps between systems because it ignores under 20ms.

- Having Back server risk, if the client time is faster than server's

3. Using NTP damon

The last way to synchronize time is using NTP daemon. The accurate daemon name is 'xntpd'

It can use the system under the different network

NTP(network time protocol) daemon is using UDP port(port 123), so you can't use if your system isn't open UPD port. Basically timed daemon immediately synchronize time after configuration but ntp daemon is synchronized after 8~10 min.(It's different by distance and drift)

- Configuration 

  1. Env: AIX 6.1

1.1 Making Server(base time) environment configuration

1) /etc/ntp.conf, /etc/ntp.drift, /etc/trace file check

2) /etc/ntp.conf file modify

   # vi /etc/ntp.conf 

==================== /etc/ntp.conf : server===============

server 127.127.1.0 prefer    # local clock as a reference, default setting

fudge 127.127.1.0 stratum 0    # values for local clock, not required

driftfile /etc/ntp.drift    # where to keep drift data

tracefile /etc/ntp.trace

=================================================

3) xntpd daemon start

#startsrc -s xntpd

xnntp daemon check command

#ps -ef|grep xntpd   
 

or

#lssrc -ls xntpd

check xntpd subsystem status is 'active' or not

1.2 Client

1) Check the gap between server and client. xntpd isn't make sync if there are over about 16min gap. so you should change the time smilary(Using # smitty date)

2) Check the below files

/etc/ntp.conf, /etc/ntp.drift, /etc/ntp.trace

3) Modify /etc/ntp.conf file 

==================== /etc/ntp.conf : Client================

server ServerIP prefer    # Input server's IP address where the ServerIP is

driftfile /etc/ntp.drift    # where to keep drift data

tracefile /etc/ntp.trace

==================================================

4) xntpd daemon start

#startsrc -s xntpd -a -X       

// -X option is prevent time backward(It controls time flow)

Check the clinet xntp daemon.

#ps -ef|grep xntpd   

 or

#lssrc -ls xntpd

If service is active normally, the subsystem's status is in 'active'

You can check the process time is synchronizing through below command
# ntpq –p

Most of all systems will be completed when the reach value is 377 but it can be long time up to time gap between server and client

Tip Usually NTP daemon is never starting when the system is reboot.
If you want to active the daemon automatically, insert the below item into /etc/rc.tcpip

start /usr/sbin/xntpd "$src_running" "-x"