HP-UX
Interview Questions and Answers- Basic and Advanced
How
to see without use of Glance the top 20 Processes in sorted order of
CPU Utilization?
Answer:
Use the command
UNIX95=
ps -e -o "user,pcpu,cpu,args,comm" |sort -rn -k2 |head -n
20
How
to see without use of Glance the top 20 Processes in sorted order of
Memory Utilization?
Answer:
Use the command
UNIX95=
ps -ef -o "user,vsz,pid,args,comm" | sort -rn –k2 | head
-n 20
How
to see the Size of a disk or a LUN in HPUX?
Answer: Command
diskinfo can be used for the same with the RAW device special file of
the Disk device to see the size of the disk /LUN as shown below.
#diskinfo
–b /dev/rdsk/cXtYdZ – Shows Size in KB
#diskinfo
–b /dev/rdisk/disk14 | awk ‘{print $1/1024/1024}’ –
Shows Disk Size in GB
How
to Mirror the root disk on Integrity Itanium Platform?
Answer:
#ioscan
– to scan the new disk say new disk
is /dev/rdsk/c0t0d0
#diskinfo
–b /dev/rdsk/c0t0d0 | awk ‘{print $1/1024/1024}’ –
shows the new disk size in GB
The
new disk should be equal to or greater than in Size than existing
root disk that is to be mirrored. The existing root disk is say
/dev/dsk/c1t1td1
See
the sizes of the 3 partitions on the existing disk c1t1d1 in MB
#diskinfo
–b /dev/rdsk/ct1t1d1s1 | awk ‘{print $1/1024}’
#diskinfo
–b /dev/rdsk/ct1t1d1s2 | awk ‘{print $1/1024}’
#diskinfo
–b /dev/rdsk/ct1t1d1s3 | awk ‘{print $1/1024}’
Let s1
size be 400MB and s3 Size be 500MB
Create
the /tmp/pdf file like this
3
EFI
400MB
HPUX
100%
HPSP
500MB
Write
the partition information to the new disk
#idisk
–wf /tmp/pdf /dev/rdsk/c0t0d0 – to create the partitions on the
new disk
#insf
–e –H< hardware_path_of_new_disk> :
install DSF for the new partitions
#ioscan
–H< hardware_path_of_new_disk> --
confirm that all the 3 partitions are created as
/dev/[r]dsk/c1t1d1s1(EFI), /dev/[r]dsk/c1t1d1s2 (HPUX),
/dev/[r]dsk/c1t1d1s3 (HPSP)
#mkboot
–e –l /dev/rdsk/c1t1td1 : This
formats the EFI partition and populates with the EFI Utilities on it
Create
a temp file called /tmp/auto with the contents in a single line such
that
#cat
/tmp/auto – shows “boot vmunix
–lq”
Put
the /tmp/auto file to new disk in EFI partition
#efi_cp
–d /dev/rdsk/c1t1d1s1 /tmp/auto /efi/hpux/auto
Pvcreate
on the S2 to assign that as bootable LVM PV
#pvcreate
–B /dev/rdsk/c1t1d1s2
Extend
the root VG to include the /dev/dsk/c1t1d1s2
#vgextend
/dev/vg00 /dev/dsk/c1t1d1s2
Mirror
all the LVs on existing root disk to the new disk s2 partition
#for
i in 0 1 2 3 4 5 6 7 8 9
>do
>lvextend
–m 1 $i /dev/dsk/c1t1d1s2
>done
Edit
the /stand/bootconf file to include the definition of new Boot LV:
append a line like below to the file and save and exit
“l
/dev/dsk/c1t1d1s2”
The
above statement has first character as small L and not number 1 (one)
Update
the LVM Boot information
#lvlnboot
–vR
Confirm
the LVM boot information is updated
#lvlnboot
–v
Confirm
that there are no stale extents in the rootvg
#for
i in 1 2 3 4 5 6 7 8 9
>do
>lvdisplay
–v $i | grep –I stale | wc –l
>done
All
Reading should be zero.
How
to boot an Integrity System to Single user mode
Answer:
At the
time of reboot of the server, attach to the console of the server and
go to EFI Menu. Enter the choice EFI built in shell to go to EFI
shell.
From
the EFI Shell type the command'
EFI
Shell> fs0:
Fs0:>
cd /efi/hpux
Fs0:>
hpux
Interrupt
autoboot here by pressing any key and you will come to HPUX>
Prompt
From
HPUX prompt give the below command
HPUX>
boot vmunix –is
How
to boot an Integrity System to LVM Maintenance Mode?
Answer:
The
approach is the same as booting in Single user mode but the option
instead “-is” would be “-lm”
How
to make syslog send logs to a remote syslog server?
Answer:
The
syslog configuration file in HPUX is /etc/syslog.conf and has the
keywords and values separated by a< TAB> space. To see the TAB
spaces do cat –t /etc/syslog.conf
To
enable a system to send all system related logs by syslog to a remote
server whose IP is A.B.C.D edit the /etc/syslog.conf to put an entry
like this
*.* @A.B.C.D
Please
note that there is a TAB between *.* and @A.B.C.D
After
this restart syslog or refresh syslogd.
Restart of syslogd is done using /sbin/init.d/syslog
stop; /sbin/init.d/syslog start
Refresh of syslogd is done using kill
–HUP `cat /var/run/syslog.pid`
How
can you test if the Syslog is logging messages to
/var/adm/syslog/syslog.log?
Answer:
This can be tested using sending a small logger message to syslogd
using the command shown as below.
#logger
–t hi hello
Logger
sends a message to syslogd and if syslog is working properly a line
should get added to /var/adm/syslog/syslog.log as showing “hi:
hello”
If
PRM is enabled and configured, how to display in PS command which
PSET group a process belongs to?
This
can be seen using the option –z of the ps command, which with other
things being displayed would also show the PSET (processor SET) group
in which the process is executing.
#
ps -fzu haydes
UID
PSET PID PPID C STIME TTY TIME COMMAND
haydes
0 20320 1 220 19:28:33 ? 6:23 itDB
haydes
1 20321 1 255 19:28:33 ? 6:45
quarryDB
haydes 0 20319 1 0 19:28:33 ? 0:05 comp-it3
haydes
1 20323 1 255 19:28:33 ? 6:43 rdr_quarry
How
to start a process in a particular PRM group and how to make the
process move to a different PRM group?
A user
can start his process or move his started process to a PRM group to
which he belongs to as per the PRM configuration file.
If the
user has an access to a PRM group, he can start his process in that
PRM group using the command prmrun as shown
$prmrun
–g< PRM_Group> <Application name>< Application
Options>
If he
is a member of the other PRM group, then he can also move his running
or started process to another PRM group with prmmove command as
below.
$prmmove
–g< PRM_Group> -p <Process_ID> | -g< Process Group
ID> | -u< User ID>
How
to quickly form a WLM configuration file when a PRM configuration is
already present?
The
default PRM configuration file is /etc/prmconf. Once a fully
configured PRM file is already present on the system, the WLM
configuration file can be generated using
#wlmprmconf
/etc/prmconf /etc/wlmconf
This
generates the corresponding WLM configuration file as per the PRM
definitions in PRM configuration file /etc/prmconf. Only to have an
instantly usable WLM configuration file, the /etc/prmconf needs to be
got rid of PRM_SYS group before running the conversion using
wlmprmconf.
How
do you differentiate the files /etc/lvmtab and /etc/lvmtab_p? And
how to read lvmtab files?
Answer:
/etc/lvmtab
and /etc/lvmtab_p in HP-UX are binary files and they can be read
using the strings command like
strings
/etc/lvmtab
strings
/etc/lvmtab_p
the
above show the Volume groups by name and the member PVs (physical
volumes) of that VG.
While
/etc/lvmtab keeps the entries for VG version 1.0 Volume Groups,
/etc/lvmtab_p contains the entries for Version 2.x Volume Groups.
Version 2.X volume groups had been introduced in HP-UX 11.31 that is
11iv3.
Question:
Which are the RAID Levels supported by LVM?
Answer: LVM
makes use of software RAIDs, RAID levels supported are
RAID 0
(Striping) | lvcreate using -I and -i options
RAID 1
(Mirroring) | lvcreate or lvextend using -m options
RAID
10 (Distributed Extent Based Striping) | using lvcreate /lvextend
with -D y -s g and with use of /etc/lvmpvg
RAID
01 (Mirror of Stripes) | Using lvcreate -m with -I and -i options (
0n 11.23 this requires the presence of the 16TB LV + Stripe mirror
patch )
Creation
of RAID 01 on HP-UX which requires the above said patch on HP-UX
11iv2 and is supported by Default on HP-UX 11iv3 with the new
generation mass storage stack is as follows.
lvcreate
-m 1 -I <Stripe_UNIT_Size> -i <number_of_stripes> -L
<LV_SIZE_in_MB> /dev/<VG_Name>
Question:
What are the Commands to configure RAID 1 in LVM?
lvcreate
-m 1 -n <LV_name> -L <LV Size in MB> <VG _Name> [
PV_names ]
If
there is an existing LV which you need to mirror then you can use the
LVEXTEND Command as follows
lvextend
-m 1 /dev/<VG_NAME>/<LV_NAME> [PV_NAME]
If you
specify the PV name, the mirrored extents will go to the specified PV
or the VG will choose the Best PV as per the Strict policy of the
existing LV.
Question:
What are the various types of strict mirroring in HP-UX?
Answer:
There are 2 types of strict mirroring in HP-UX LVM which are PV
Strict mirroring and the PVG Strict mirroring.
PV
Strict mirroring is specified using the -s y flag of lvcreate and
lvextend.
PVG
Strict mirroring requires the -D y -s g options and also uses the
/etc/lvmpvg definitions.
Question:
What is the LVMPVG files used for? How is this used for Extent based Striping and mirroring? What is DEBS (Distributed Extent Based
Striping)
Answer:
LVMPVG file /etc/lvmpvg is a text file which can be created anytime
on a system and specifies that for a given VG which are the PVs which
are part of the Physical Volume groups.
A
sample /etc/lvmpvg can be like this.
/etc/lvmpvg
VG
/dev/vg01
PVG
pvg01
/dev/dsk/c0t0d0
/dev/dsk/c0t0d1
PVG
pvg02
/dev/dsk/c1t0d0
/dev/dsk/c1t0d0
The
above file shows that the pvg0 in the VG vg01 come from controller c0
whereas the PVG pvg02 comes from the disks at the controller c1.
We
will create an LV using the pvg01 disks and then will mirror the same
with the disks in the PVG pvg02. The below is also known as
distributed extent based striping.
lvcreate
-n my_lvol -L 5000 -D y -s g -g pvg01 /dev/vg01
The
above creates an LV of size 5000MB with name my_lvol in the VG vg01
and in the PVG pvg01 which means the disks /dev/vg01/c0t0d0 and
/dev/vg01/c0t0d1 with the (-D y) which is the Round Robin
distribution of the accross the PVs in the PVG pvg01. Also -s g tells
LVM that the LV has mirroring policy as PVG-Strict.
In the
next step we mirror the LV
lvextend
-m 1 /dev/vg01/my_lvol -g pvg01 /dev/vg01
Which
creates the mirror of the LV to the disks in pvg01 that is the mirror
goes to the disks /dev/dsk/c1t0d0 and /dev/dsk/c1t0d1 because of the
PVG-strict policy which ensures that the mirrors of the PEs cannot
remain in the same PVG.
The
above is also called Distributed extent based striping and mirror
which is an example of RAID 10. Also in case of DEBS the Stripe size
as evident is equal to the size of the Physical extent.
Question:
How to create a striped LV on HP-UX?
Answer:
The creation of striped LV can be done as below.
lvcreate
-L <LV_SIZE_IN_MB> -I <Stripe_UNIT_SIZE> -i
<number_of_stripes> -n <LV_NAME> /dev/<VG_NAME>
Question
: How to create a spanned Logical Volume?
Answer:
The Spanned logical volume gets the Physical Extents from the
multiple PVs in a VG. This can also be called to be similar as
implementation of the Concatenated volume.
The
creation can be done as follows.
lvcreate
-n my_spanned_lv /dev/vg01
This
creates an LV of 0 size in the VG vg01 with the LV Name as
my_spanned_lv.
Extend
this LV to the disks of the VGs as shown.
Assuming
that the VG contains three PVs as /dev/dsk/c6t0d0 /dev/dsk/c6t0d1
and /dev/dsk/c6t0d2 and I want 333 MB coming from each disk, we can
do the following.
lvextend
-L 333 /dev/vg01/my_spanned_lv /dev/dsk/c6t0d0
lvextend
-L 333 /dev/vg01/my_spanned_lv /dev/dsk/c6t0d1
lvextend
-L 333 /dev/vg01/my_spanned_lv /dev/dsk/c6t0d2
Question:
How does LVM decide which disk (PV) to use to allocate extents of
the LVs being created when no specific PVs are specified.
Answer:
When a VG is created, the PV instances get an order or sequence
number in the order the PVs are added to the VG. This sequence number
is known to the LVM for each VG. So in a VG when an LV is created or extended without any specific device reservations, the order in which the Physical extents are allocated is the same as the sequence
number as kept by LVM for the VG.
The extent start getting allocated from the first PV first. When the
free PEs on the first PV are exhausted that goes to the second PV and
so on.
In
case you have made an LV as PV strict that will mean that the PEs
which are mirrored for that LVs will strictly reside in the different
PVs and LVM ensures to maintain that.
So if
you are creating the LV, the first sequenced PVs is used to create
the LV. Then when the mirror is created, the extents come from the
next PV even if PV1 has free PEs. This happens as a combined result
of the PV strict mirror policy and the Sequence of the PVs (the order
in which the PVs had been added to the VG) information which LVM
maintains and for the fact that there has been no device reservation
specified either for the original PEs of the LV or for the mirror PEs
forthe original PEs.
How
to identify if a Logical Volume is mirrored or not?
You
can you lvdisplay command for the LV like
lvdisplay
<LV_Name>
See
for Mirror Copies value, if it is 1 : this is a 1 way mirror if 0 the
LV is not mirrored.
For a
1 way mirrored LV, the value of the Physical extents is just twice
the number of the Logical Extents as each LE maps to 2 PEs which are
in mirror relation.
In
case there is no mirror for the LV you can see the counts of PE and
LE to be equal meaning that 1 LE corresponds to 1 and only 1 PE.
How
to check if the mirroring has an issue? How to check the health of
an LV which has mirror?
In
case there are issues with the mirroring you can see the Stale PEs on
the LV.
To see
this you can use
lvdisplay
-v <LV_Name> | grep -i stale
If
mirror copies in lvdisplay is 1 and there are no stale extents you
are good, your 1 way mirror is healthy.
How
to break the mirror and change root disk?
Breaking
mirror.
Need
to make sure first that we know which mirror to break.
I have
root disks mirrored in which /dev/dsk/c0t0d0 is mirrored to the disk
/dev/dsk/c1t1d1 and I want to replace the disk /dev/dsk/c1t1td1 by
the disk /dev/dsk/c2t2d2 which is a free disk ann not being used and
is same or bigger in size than /dev/dsk/c1t1d1
To
remove the Mirror Extents from the disk /dev/dsk/c1t1d1,
for i
in /dev/vg00/lv*
do
lvreduce
-m 0 $i /dev/dsk/c1t1d1
done
lvlnboot
-vR : update the LVLNBOOT information.
vgreduce
/dev/vg00 /dev/dsk/c1t1d1
The
above removes the disk from the VG roovg vg00
Add
the new disk and mirror the LVs to the new disk.
The
below is for a PA-RISC system, for Itanium Systems there is an EFI
partition to be present on the root disk and the process is slightly
different.
ioscan
-fnC disk
diskinfo
-b /dev/rdsk/c2t2d2 : verify
the size of the new disk to which mirroring will go.
pvcreate
-B /dev/rdsk/c2t2d2
mkboot
-l /dev/rdsk/c2t2d2
mkboot
-a "hpux -lq" /dev/rdsk/c2t2d2
vgextend
/dev/vg00 /dev/dsk/c2t2d2
for
i in /dev/vg00/lv*
do
lvextend
-m 1 $i /dev/dsk/c2t2d2
done
The
above may take substantial time.
Once
the mirror completes update the lvlnboot and /stand/bootconf
lvlnboot
-vR
Edit
the line pertaining to older /dev/dsk/c1t1d1 and replace that with
/dev/dsk/c2t2d2
How
to see the default route in HP-UX? How to add default route in HP-UX
?
You
can make the use of the command netstat -rn and look out for the line
having "default" and the Routing Flags as UG.
To add
the default route you can do as below.
Assuming
that the LAN interfaces are configured with IP address and the Subnet
mask as suggested by the Network team as per the design of network at
your place ..
To add
the Route temporarily you can
1)
First ping the gateway ( I assume the default GW to be set is
10.1.1.1)
ping
10.1.1.1
2) If
the ping is successful ( Assuming that ICMP is not blocked from your
server to the router interface) you can perform below to add the
default route (gateway) temporarily.
route
add net default 10.1.1.1 1
The
last 1 is called the metric and is 1 for immediately connected route
of the network.
To add
the default gateway permanently you can make use of SAM and also that
of set_parms where in you can configure the network parameters for
your system.
How
to see all the LAN interfaces in HP-UX? How to see MAC addresses of
all the LAN interfaces?
To see
all the LAN interfaces in HP-UX inclusive of all those are configured
as well as those which are not configured use the commands as below.
ioscan
-fnC lan
or
ioscan
-kfnC lan
lanscan
lanscan
will show the MAC addresses for all the interfaces also.
How
to take OS backup? How to take Ignite backups of the HP-UX system on
to a Tape drive?
First
identify if your system has a tape drive in place.
ioscan
-fnC tape
or
ioscan
-kfnC tape
Say
the tape device is /dev/rmt/0m and has the no-rewind device file for
the same tape as /dev/rmt/0mn
To
make the Tape archive of the system use the no rewind device that is
/dev/rmt/0mn.
Load
the tape media in the tape device.
Confirm
that the media is loaded and in case needed rewind the tape to the
home position.
mt
-t /dev/rmt/0mn status
To
rewind to the original tape home position you can do
mt
-t /dev/rmt/0mn rewind
or
in
short
mt
-t /dev/rmt/0mn rew
Confirm
the tape is positioned
mt
-t /dev/rmt/0mn status
To
make a Tape archive backup you can use the command as
make_tape_recovery
-Av
This
automatically takes backup to the default tape drive configured on
the system.
make_tape_recovery
-v -a /dev/rmt/0mn -x include_entire=vg00 -x exclude=/var/adm/crash
The
above command takes backup of the system into tape drive /dev/rmt/0mn
and includes entire VG vg00 (which is included in Ignite backup by
default) in -v (verbose mode) and excludes the Directory
/var/adm/crash.
To
include the other VGs apart from the rootvg vg00, the command becomes
like this.
make_tape_recovery
-v -a /dev/rmt/0mn -x inc_entire=vg00 -x inc_entire=vg02 -x
exclude=/var/adm/crash
The
above command additionally takes the vg02 data also in the Ignite
tape backup.
How
to take the network Ignite-UX image backup for a system? How to take
make_net_recovery backup?
Assuming
that you have an Ignite-UX server in your environment, you can use
the make_net_recovery command.
Some
of the invocations of make_net_recovery are as below.
These
commands are executed from the HP-UX server which is a client to the
Ignite-UX server.
make_net_recovery
-v -s <Ignite-UX_server_name> -x inc_entire=vg00
The
above takes a verbose mode Ignite-UX make_net_recovery backup to the
Ignite sever. It os noteworthy that by default the make_net_recovery
archive of the client is put at the
/var/opt/ignite/recovery/archives/<hostname> directory on the
Ignite-UX server, where <hostname> is the client name for which
you are making an archive in make_net_recovery format.
The
above also updates the client information for the recent backup being
taken for the client on the clients folder of the Ignite-UX server,
the location on the Ignite-UX server being
/var/opt/ignite/clients/<hostname>. Where hostname is the name
of the client which you are creating a network make_net_recovery
archive for.
make_net_recovery
-v -s <Ignite-UX_server> -a <Archive_server>:<Archive_location>
-x inc_entire=vg00 -x exclude=/var/adm/crash
In the
above case my Ignite-UX server is different from the Network archive
server and I want to exclude /var/adm/crash of the client from being
backed up.
To be
noted that the Ignite-UX backup takes help of NFS in the background
where the Ignite-UX server and the archive server (in case in your
environment the Ignite-UX and the archive servers are different) work
as NFS server and the client is an NFS client.
This
mandates that the folders /var/opt/ignite/recovery/archives and
/var/opt/ignite/clients are NFS exported for read/write and access
from the client which is being backed up.
To be
more specific you can give access to the Ignite-UX clients in a more
restrictive manner by exporting the individual
/var/opt/ignite/recovery/archives/<hostname> and
/var/opt/ignite/clients/<hostname> to the specific client
(hostname which is being backed up using make_net_recovery) something
like below.
/etc/exports
on Ignite-UX server
/var/opt/ignite/clients/host1
-anon=2,access=host1.example.com:host1
/var/opt/ignite/recovery/archives/host1
-anon=2,access=host1.example.com:host1
Please
note that if FQDN is used in your environment use of the FQDN as
above helps to overcome some permission denied error messages.
-anon=2
above means that during the backup the userid "bin" is
being used for NFS mounts to mount and write to the
/var/opt/ignite/clients/<hostname> and
/var/opt/ignite/recovery/archives/<hostname>
