HPUX Basic Doubts

[sudhapage]

Hi all,

I have some basic doubts in HPUX. Please clarify my below doubts.

1. Crash Dumps - What is the use of Crash Dumps? How we can use it?

2. How to increment PATH variables.

Ex: If I want to add one directory value with existing value for PATH Variable?

3. What abt tombstones? How we can use it?

4. Is there any documents for HPUX Installation with snapshots?

5. what is the difference between ssh and telnet?

6. If temperature getting very high, server will through any error messages?

7. If one hard disk is failed, how we can find it manually in server? Once I heard, using dd command we can able to find it. If yes how to use this command to find out the harddisk.

8.What is the use of GSP?

9. How much size we should give for swap space? 1.5 % or 2 % ?

Regards,
Sudhakaran.K

[inventsekar_1]

1. Crash Dumps - What is the use of Crash Dumps? How we can use it?

Crash dumps contains information about what caused the crash.

2. How to increment PATH variables

# export PATH=$PATH:/opt/test
will add the /opt/test in the existing PATH.

3.
4. without snopshots only available from hp website. http://docs.hp.com/en/oshpux11iv2.html#Installing%20and%20Updating
if u still want a guide with snopshots, reply ur email id. i will send u.
5. SSH is more secure than telnet. name itself "secure sh(shell)". other than security, no difference between ssh and telnet. quick search in google will give lot information.
6. yes. i think.
7.
8. http://en.wikipedia.org/wiki/User_talk:Gsp
9. twice the phical RAM size is better. also it depends on the applications running on the system.

[sudhapage]

hi
** edited to comply with forum guidelines **

Anyone please answer for me for my 3 & 7 questions & I need more explanation for GSP & Crash Dumps.

Regards,
Sudhakaran.K

[inventsekar_1]

Crash dumps:
man crashutil

and "Analyzing crash dumps is not a trivial task. It requires intimate
knowledge of HP-UX internals and the use of debuggers. It is beyond the
scope of this document to cover the actual analysis process. If you need
help analyzing a crash dump, contact your Hewlett-Packard representative."

http://docs.hp.com/en/B2355-90950/ch05s05.html

this contains good information about crash dumps.

[Baiju Kumar.B_1]

1)crash dumps ---crash dump is a dump it will creating due to the system crash.by using some special analizing tool ,it is possible to find out the cause of system crash
2)export PATH=$PATH:(required directory)
eg: export PATh=$PATH:/home/sudhakaran
3)tombestones is a processor log .it is under /etc/tombstones directories
4)you can find out from www.docs.hp.com.there you will get the proper installation orocedure
5)ssh is offer more security than telnet for remote connections
7)use ioscan -funC disk command in hp ux
if you will get the open message in any of the hard disk's place it will be faulty
other option is diskinfo /dev/rdsk/.
or #dd if=/dev/rdsk/ of=/dev/null bs=256 if you are getting the shell prompt at last no problem in hard disk if some read error is coming there will be problem in hard disk after diag check the hardware path of hard disk and remove the faulty disk
6)if it is critical error you will get the mail to root by event monitoring service(on line diag tool)

8)it is a service processor login other than normal login.youcan use this login for system diagnostics purpose,suppose during system get hang and you can able to check chasis logs also
9)it is either two or theree times of the physical memeory of the system.Refer the application mannual for get the proper swap space required for applications

regards
Baiju kumar.B

[Yogeeraj_1]

Hi Sudhakaran,

Gaurdian Service Processor (GSP)
GSP provides you the single point administrartion. you can access the consoles of different machines(similar to web console).

Below a quote the i have recorded in my notes from Bill Hassell:
The GSP is found on N,L,R,A-class machines as well SuperDome. The GSP is primarily for hardware issues and normally the system administrator never uses the GSP. The GSP logs hardware and console messages in special NVRAM on the processor.

You can access the GSP with the console using CTRL-B, or through the special GSP LAN port. Or you can use an HP Web Console adapter connected to the RS-232 console port to use a web page to look at the console. This is useful since it allows you to remotely interact with the hardware when HP-UX is not running, for example, booting interactively.


hope this helps too!

kind regards
yogeeraj

[Torsten.]

Hi,

some additions:

6) you will get overtemp critical or emergency messages. If a critical value is reached, the system will do a "reboot -h" (not a shutdown). If the temp is getting even higher, the GSP will power of the system. (see "man envd" for more information)

7) this could be hard. You have to know your hardware equipment very well. Each slot inside the server is associated with a hardware path, you need to be able to "read" the path.

example:

disk 3 0/2/1/0.8.0 sdisk CLAIMED DEVICE HP 73.4GMAW307
3NC
/dev/dsk/c4t8d0 /dev/rdsk/c4t8d0


This is a disk connected to a rx1600.
0/2/1 is a dual SCSI controller in the lower PCI slot
0.8.0 is the first slot in the connected DS2100

You see, you need to know your hardware well.

8) The GSP take care of your server and is a tool to configure it. The functionality is slightly different between different server models.
Newer service processors are also named MP / iLO / MP iLO 2

9) at least the same amount as your memory is to be able to hold your dump

[A. Clay Stephenson]

9) The correct answer is as much as you need. The old rules about 2x-3x RAM for swapspace very seldom apply anymore and haven't applied for decades. I have many machines with much less swap than memory and they run very well.

[A. Clay Stephenson]

9) Swap and dump should have absolutely nothing to do with each other; defining separate dump space also means that there is no reason to save and compress the dump before swap is activated; moreover, dump should never be mirrored and swap should always be mirrored.

[Jan van den Ende]

Sudhakaran,


What is the use of Crash Dumps? How we can use it?

Like Sekar wrote:

Analyzing crash dumps is not a trivial task

Have I ever seen an understatement!
Before me is the certificate that I have completed the "Crash Dump Analysis Course", so now I have "the basic tools to investigate crash dumps"
That course was a VERY INTENSIVE 60 hour course!
But it DID teach the fundamentals of how to trace back through a dump, and pinpoint EXACTLY which piece of programm code encountered which condition leading to the crash. And since only privileged code can cause a crash, usually that gives you the precise ammo to contact the supplier of that piece of software. (unprived code should not crash the system, only the process running it).

All in all, the part of this that is the concern of the SysAdmin is the need to HAVE a dump when the system has crashed, and to transfer that dump to the HP specialists.

hth,

Proost.

Have one on me.

jpe

[Steven E. Protter]

Shalom,

The telnet versus ssh discussion brings up a general security concern.

The real difference between telnet and ssh is that telnet authenticates in clear text as does ftp which can be replaced by sftp/scp which is part of the same package as ssh.

It is possible to use kerberos to authenticate ftp and telnet, I did it once at an HP Internet Security class. The problem is that most common ftp and telnet clients wouldn't know what to do with such a server, making that practice of questionable value.

If in fact you are new and building a new server see if you can get away with telnet and ftp disabled, replaced by openssh which is free on the hpux side from http://software.hp.com Search for Secure Shell.

SEP

[Bill Hassell]

> 6. If temperature getting very high, server will through any error messages?

I always get worried when I see this type of question because companies will spend hundreds of thousands of dollars on disks, tapes, network equipment and computers, but spend virtually nothing on proper airconditioning with multiple backup solutions. If the question has come up, it probably means you already have a problem. Now it's important to note that only newer models of PARISC and Integrity/Itanium computers will actually shutdown at high temperature. Older systems will do exactly what your other equipment does -- run until it completely fails.

To me, it makes no sense to purchase any computers and network components if they are put into an environment that can destroy them in a few minutes. Note that unless you have staff in the computer room day and night including weekends, you should shutdown the systems when no one is there. An airconditioning failure will damage the equipment long before anyone can get to the building and shut them down.

[A. Clay Stephenson]

6) Who cares? The correct approach is to have N + 1 HVAC units where N is the number required to cool the data center --- and if your computers are generator backed then your HVAC units must be battery backed. Ideally there is also a thermostatic breaker that will interrupt power to the entire data center during a temperature excursion. I would never rely upon any computer to alarm upon temperature or shut itself down. Even if the computer is smart enough to shutdown, most disk drives just keep running.

Computers are terrible instruments with which to measure temperature but there is an instrument that does measure termperature well; it's called a thermometer. Find one with a serial or network interface and use it to monitor your termperature but always have N + 1 units because I wouldn't trust anything (except a thermostatic breaker) to prevent damage.

[Emil Velez]

8. Use of GSP.. If you use high end PCs with RILO cards it provides the same plus more functionality.

7. If a hard disk fails and it is part of a mirror LV then there will be syslog messages and EMS alarms that can be configured when that happens. If you are like many people and use disk arrays (yes they can be used for boot disks too) you never have a disk fail (then you depend on disk array detection and replacement procedures)

4. Installation with snapshots probably mean using ignite to create a snapshot of a system that can be easily restored. look up a command called make_recovery or the ignite users guide.

1. Crash dumps is for kernel panics. If your system crashes it saves the state of memory and if there was a hardware or software defect it can be found after the system boots and you send the crash to HP.

5. SSH vs telnet is not a HPUX specific issue all UNIX systems linux,bsd, solaris, aix etc support these network services.

I hope these answers help..

Emil

[sudhapage]

Hi all,

Really Thanks for all of your responses.

Anyway I am not clear with above 2 questions!

one is tombstone. Tankkapan you have mentioned tombestones is a processor log. It means its will write the log when the time of failure or after removing any one of the processor in system? or Ususally it will write the status about processor?

second one is physically find the faulty harddisk? I understand now, without familiar about hardware path we can't do this one. In D or K class servers having any visual informations about hardware path? Like stickers or etc...

Regards,
Sudhakaran.K

[Victor Fridyev]

Hi,

A tombstone is a list of register values used for troubleshooting if computer failed.
http://docs.hp.com/en/A3725-96005/ch07s04.html

D and K class servers have stickers with HW addresses on their cover(just open cover door), so you can find needed information.


HTH

[nanan]

Hi
Tombstone is useful when the system encounter an abnormal HW failure causing system crash.

It can be analyzed to find the root cause of the problem
It is related to HW not SW
when SW raise a problem, system get panic and dump current mem.
and it will save to crash directory


HDD fault,
thess day! most HDD having LCD and you can see the physical status.

There are two cases
1. Using Enterprise Storage system
you don't have to try where the faulty disk be, Storage system can do

2. Using local disk and local enclosure
It is not difficult to find out a faulty disk
HP provide the menual of HW Path
and you can know PCI IO paths
Be familier with IOSCAN command


Regards
nanan

[Dineshkumar · Accepted solution]

Sudha,


1. Crash Dumps - What is the use of Crash Dumps? How we can use it?

Crash dump is to find out the root cause of system panic (mostly software panic). If itâ s enabled you will be able to find out the trace why system got panic (This can be done by an HP representative) i.e Path stored /var/adm/crash/

2. How to increment PATH variables.

export PATH=$PATH:(required directory)

3. What abt tombstones? How we can use it?

Tombstones will help to find the root cause of hardware panic (This can be done by an HP representative) i.e Path stored /var/tombstones/ts99 â latest

4. Is there any documents for HPUX Installation with snapshots?

This question was not clear , please re-phrase are you referring to file system like vxvm & JFS if so please refer to admin guide of the product

5. what is the difference between ssh and telnet?

Telnet â unsecured connection, less advantage
SSH â Its secured connection â more advantage like user can have public & private keys to access the server and supports different authentication techniques
Supports scp â secure copy sftp â secure ftp

6. If temperature getting very high, server will through any error messages?

Temperature issues are been monitor by EMS (Event Management services)
Stored at /var/opt/resmon/log/event.log

7. If one hard disk is failed, how we can find it manually in server? Once I heard, using dd command we can able to find it. If yes how to use this command to find out the harddisk.

You can refer this file /var/opt/resmon/log/event.log and find the hardware path of the disk like 0/5/0/0.2.0 and number of time the events has been reported

diskinfo â v /dev/rdsk/

dd if=/dev/dsk/cxtxdx of=/dev/null

Then you can cross check with ioscan no_hw mostly its disk issue needs to replace

Mostly scsi disk write errors would appear in syslog.log

8.What is the use of GSP?

To put in a simple way, its mini system to manage the enter system which function as standalone system with its own processor and network etc.,

9. How much size we should give for swap space? 1.5 % or 2 % ?

Its recommended to have swap as 2 % of the memory

[John_Hancock]

In addition to the above...

5. As well as the security implications SSH has far more facility that telnet. Shared key authentication - ie login without password request, remote copy, remote execute all secure encrypted.

7. "ioscan fnC disk" of course but EMS is an HP supplied utility that is installed by default which performs hardware monitoring and will notify via syslog and email in the event of a failure.

8. The GSP is a seperate processor (Guardian Service Processor) that will allow you to manage your system as long as there is power to the system. You can do such things as console access, set the default boot path, boot from alternate path, power down, power up and power cycle the system. In older systems this was serial (RS-232C only), but in later systems you have serisl, lan and web interface. However we found the lan interface to be the most convinient.