extending a partition on Linux / VMware /SAN disk without gparted

First of all extend the hard disk size, either using vmware console, san console or vmdisk-vmanager command.

In my example i extend sda from 10 to 12 GB.

In my current linux:
/dev/sda1,2,3 --> primary partition

/dev/sda4 --> extended partition
/dev/sda5 --> inside sda4, it is also the /home partition

After extend, see the result using fdisk -l,
Notice if the HD size now is 12 GB.
However the partition sector/unit is still remain the same as before.


Since the last sector is /dev/sda4, so now we extend it sda4 first, then we could extend the sda5.

using parted :
# select /dev/sda
# unit mb
# print
# resize 4 xxxxx yyyyy

xxxx is the beginning of the sector
yyyy is the end of sector

Notice that now the /dev/sda4 has some more spaces.

after that we need to resize the /dev/sda5

Follow this link to extend the size:
http://www.hermann-uwe.de/blog/resizing-ext3-partitions-with-parted

>>>>>>>>>>>

unfortunately resize ext3 can be failed with error like "file system has an incompatibility feature" which is caused by the ext3 has been customized by redhat.

The quick way is just by using fdisk, remove the sda5 partition, and create again using the total sector of /dev/sda4.

Now u have /dev/sda5 extended size.


The longest way is by modifying all the custom attribute redhat:
http://gtechnotes.blogspot.com/2005/10/parted-gives-error-filesystem-has.html

>>>>>>>>>>>>>>

fedora 14 and intel graphic card : x11 won't start

Fatal server error
caught exception

Check your video driver:
lsmod | grep video

==> i915

Incase wanna see the card:
lspci

change the driver:
/etc/X11/xorg.conf

OR

system-config-display

(the GUI will work eventhough your X still not on)

>>>>>>>

choose VESA driver and if appear depth error message, just change the color depth to lower value.

>>>>>>>>

when u choose wrong driver, like NV (nvidia), it will give u error device not found, cause the card is not there.

multipath on linux - how the disk look like

modprobe -l | grep the "driver desc"

if not there, install it...
ex. go to brocade, download, install with some options like --update-initrd

after this the /sys/class/fc_host folder is there

after you scanning with
echo "1" > ............./issue_lip
echo "- - -" > ................../scan

the device will appear in proc/scsi/scsi file.

when you part it, you need to restart it:
ex. dm-2 will have dm-2p1 as partition 1, but you can not format it untill you reboot it, it will add another disk called dm-3 whcih is same with dm-2p1.

RHEL 4 does not have yum

You can either install it manually with some deps packages, or just use GUI :
system-config-application --isodir=/isoFolder

useful also when u have no CDRom

syslog in linux/redhat

File is in /etc/syslog.conf

by default:
*.info;mail.none;authrpiv.none;cron.none /var/log/messages


meaning logging all subject at info (and above) level but the mail and cron not.
to the messages file

To make it particular:
kern.=info (only kernel with level info)

Numerical Code Severity Reference
-------------- ------------------------------------------- ---------
0 Emergency: system is unusable [RFC3164]
1 Alert: action must be taken immediately [RFC3164]
2 Critical: critical conditions [RFC3164]
3 Error: error conditions [RFC3164]
4 Warning: warning conditions [RFC3164]
5 Notice: normal but significant condition [RFC3164]
6 Informational: informational messages [RFC3164]
7 Debug: debug-level messages [RFC3164]


7 is the most detail severity level.

Explanation free Command in Linux

total Used free shared buffers cached
Mem: 49415736 49139132 276604 0 5656 47368280
-/+ buffers/cache: 1765196 7650540
Swap: 5237180 20 5237160

Mem total : Mem used + free
Mem used : 49GB (ps + buffers 5656 + cache 47GB)


- buffer/cache : Mem used - (buffers 5656 + cache 47GB)
+ buffer/cache : Mem free + (buffers 5656 + cache 47GB)

Swap (paging/Hard disk) used is 20KB only.



Buffers and cache are considered low priority, they r using large mem when none uses it. when apps using it, they will be decreased.

Buffers is for remembering the FS like dir structure (ls -R . will boost the buffr)
cache is storing the file content, so processes can access file in cache instead read it over and over from the disk.

SCSI harddisk on linux

Manually:
cat /proc/scsi/scsi
ls /sys/class/scsi_host/.....
==> contain the host:bus:ID:Lun and Port name

By command:
lsscsi (many flag options)
fdisk -l
sginfo

Note : /dev/sdX mapped from /dev/sgX

When a HBA attach to SAN which contain 4 logical disks:
/dev/sdb,c,d,e (the partner is /dev/sdf,g,h,i)

How to see the partner :
e2label /dev/sdb XXX ==> to lable the HD
e2label /dev/sdf --> to see if it has same lable or not

Will be like this :
H:B:I:L == 5.0.0.1,2,3 (for sdb,c,d,e) and 5.0.1.1,2,3,4 (for the partner)

H is host, same as comes from one HBA. (diffrent from CDROM and LOCAL HD as it connects from different card)

B is BUS : dunno what
I is ID : same for LUNS (show same S/N of harddisk)
L is LUN : different for each partition/logical disks

additional commands:
hdparm
sdparm
scsi_id
parted /dev/sdX print ==> to see File System of a harddisk
rescan_scsi_bus.sh

Linux HD ATA problem : fsck, e2fsck

when disk has problem, normally you done fsck.

when fsck generates error like ext2 not recognized, please run e2fsck -b xxxx, this one what you must do:

1. linux rescue
2. fdisk /dev/sdX
3. fsck to each partition, NOT to the disk /dev/sda NOR to the swap NOR to the sda4(extended partition - 0 length partition)

If you got error about superblock, then:
4. each partition : mkfs -n -S /dev/sdXY
5. Get the superblock backup number xxxx, xxx, xxx and run : efsck -b xxxx /de/sdXY, try till succesfully

Reminder : Do not fsck to the disk /dev/sdX, but do it to the partition instead.

Linux WWPN and adding SAN disk

To rescan an entire SCSI bus and re-register all devices found:

When dealing with fibre attached storage, it is necessary to issue a LIP (loop initialization primitive) on the fabric:
shell> echo "1" > /sys/class/fc_host/host#/issue_lip

Replace the “#” with the number of the SCSI bus to be rescanned.

Note: Issueing a LIP (above) on Red Hat Enterprise Linux 5 is all that is needed to rescan fibre attached storage. Once the LIP is issued, the bus scan may take a few seconds to complete.

To rescan all other SCSI attached storage, a rescan should be issued:

shell> echo "- - -" > /sys/class/scsi_host/host#/scan

Replace the “#” with the number of the SCSI bus to be rescanned.

In addition to re-scanning the entire bus, a specific device can be added or deleted for some versions or Red Hat Enterprise Linux as specified below.


To remove a single existing device explicitly
shell> echo 1 > /sys/block//device/delete



OR, use the rescan-scsi-bus.sh under sg3-utils.rpm


To see the WWPN:

at BIOS : CTRL-Q Fast Util

at OS : qlogic : /proc/scsi/qlxxxx
emulex : /sys/class/scsi/device/driver/...../portid

RHEL6 - Instalation, rpm, repo

After install by default you dont have x windows:
- install xorg-x11-......
- install gnome-.........

The problem is too many packages, you must use:
yum groupinstall "X window system"
yum groupinstall "Desktop"

It used to be "GNOME Desktop Environment", but now the Desktop is refered to GNOME.

If you dont know the file related to which group, just open the comps.xml or repomd.xml file as explain further below.

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

if you find problem group not available, check by :
yum grouplist ==> to see all available group pacakes

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

if that command shows data group not available, so you must:
cp xxxxxxcompsxxxxx.xml to local repo directory.
cd to repo directory
createrepo -g xxxxxxxxcompsxxxx.xml .
yum clean all

this will build repodata folder in your local repo folder.

You copy xxxxxxcompsxxxxx.xml from the same DVD in where you copy all the rpms to your local repo folder. Or you can use the DVD of rhel5 or other version, but maybe there are discrepancies of the packages in the group install, it works though.

If the repomd is overlapped with anotehr comps.xml files, just delete the repodata folder and run the createrepo -g again.

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>


createrepo you must install using rpm (or yum) but since this means you have no repo yet, so either way you must install the 2 deps packages manually.

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

rpm --nogpgcheck

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

how to create local repo:
1. copy all the rpms from DVD to a local folder
2. in that local folder type : createrepo .
3. edit the yum.conf in /etc/yum
put the name of your repo (whatever)
baseurl=file:///folder
enabled=1
gpgcheck=0

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

rpm -qa (checking what packages installed in your PC)
rpm -q (chcking if it is insalled or not) same as above but the output diffrent though same meaning.

rpm -qi (see info of the packages)

REMEMBER : dont use * in the packages name, use : rpm -q | grep XXXX instead.
Note : if your CWD has pacakges files also, it cant be confusing, be careful!!

>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>

VMware : ESX vs ESXi vs ESXiEmbedded

ESX has service console, service console is restricted shell of redhat OS

ESXi is not using service console, u have to use RemoteCLI to manage it, but still ESXi needs a Harddisk to store. The donwside is could be hard to troubleshoot for newbie since not using service console.


ESXiEmbedded is ESXi but stored in flashdisk instead, the downside is flash disk has no mirroring function, when it is broken, u need to install new one.