dmesg

[Uday_S_Ankolekar]

dmesg is not showing memory information. Is there anyway I can clear circular buffer so I can get proper dmesg output. No I don't want do a reboot

OS:HP-UX 11.x on a N box

[Andreas Voss]

Hi,

have a look at the file

/var/adm/syslog/syslog.log

All system buffer messages are kept in this file.

Regards

[James R. Ferguson]

Hi Uday:

My favorites for obtaining memory are these:

# grep -i physical grep -i physical /var/adm/syslog/OLDsyslog.log

# echo "selclass qualifier memory;info;wait;infolog"|cstm > /tmp/meminfo

...JRF...

[eran maor]

Hi

you can also get the info of the memory with the adb command :

For 10.X execute:
echo "physmem /D" | adb /stand/vmunix /dev/kmem

This should return:

physmem:
physmem: 16384

For 11.X 32 bit execute:
echo "phys_mem_pages /D" | adb /stand/vmunix /dev/kmem

This should return:

phys_mem_pages:
phys_mem_pages: 106496

[James R. Ferguson]

Hi again:

Ooops! One 'grep' will do... :-))

# grep -i physical /var/adm/syslog/OLDsyslog.log

...JRF...

[A. Clay Stephenson]

Hi Uday,

No, the buffer is circular and when it overlaps it's gone. But a better way to do what you want to do is:

echo "phys_mem_pages /D" | adb /stand/vmunix /dev/kmem

This will give you the amount of memory in 4096 byte pages.

Clay

[MANOJ SRIVASTAVA]

Hi Uday

there are 2 things you want to do

1. Check memory Size , you can either do it throught the dmesg | grep Phy or there are other ways to do like quoted by others .Or got to SAM--->Performance Monitor---> System Properties .

2. Set the dmesg output correct .In this case if you are not getting dmesg bcoz of

The current version of the kernel doesn't match the current version of the symbol table symtab in /stand/dlkm. This can result if kmupdate is not run after building a new kernel.

then may be u will have to rebuild kernel with no reboot option.

Manoj Srivastava

[Bill Hassell · Accepted solution]

One additional note is to keep a dmesg log so the limited buffer size will not be a factor. This is shown in the /usr/newconfig directory where original copies of config files are located:

/usr/newconfig/var/spool/cron/crontab.root

You'll see the lines:

# log kernel diagnostic messages every 10 minutes
05,15,25,35,45,55 * * * * /usr/sbin/dmesg - >>/var/adm/messages

Change the name of the logfile from messages to dmesg.log--it is much more intuitive. dmesg has this cool option: dmesg - which means: show only that which has changed since the last time dmesg - was run.

Thus, the above command does nothing most of the time until something is written to the console and dmesg - adds the information. Plus dmesg - adds a time stamp at the beginning of the new information. Now you won't have to worry about dmesg overflowing.

[Uday_S_Ankolekar]

Thanks all for the responses.
My syslog file shows this error
vmunix: lvm: pv3 has been returned to [vg1]
vmunix: lvm: pv1 has been returned to [vg1]
This is the same output I get when I run dmesg.
What is this mean..??
I checked with vgdisplay, pvdisplay nothing seems to be wrong .

[Peggy Fong]

Hi

You probably have several pvs (physical volumes or disks) in vg1 and pv1 and pv3 (the first disk and third disk in the volume group were temporarily unavailable. The message in syslog and dmesg indicates that they are now available - started responding which is why everything looks ok now. Usually there will be a earlier message indicating a problem - disk unavailable or similar message.

HTH
Peggy

[A. Clay Stephenson]

Hi,

This indicates that you are having SCSI disk problems. Since these are occuring on different drives, I suspect that you have a cabling, controller, or terminator issue. You can look in /var/arm/syslog/syslog.log for more detailed messages. This can also be caused by device timeout errors and you may need to increase the timeouts using pvchange (especially if this is a disk array). The system is recovering from these errors (at least for now) but you should fix this before it becomes more serious.

Clay

[Uday_S_Ankolekar]

Paggy,
Syslog don't have any such error What else should I check??
This message keeps on coming in syslog dmesg also gives the similar output. How do I stop this?

[Bill McNAMARA_1]

relating to the lvm errors,

ll /dev/*/group
the volumegroup with the group file entry 0x010000
let's say for example it ends up to be vg01

issue a strings /etc/lvmtab

Lets say it looks like this:
.
.
.
/dev/vg01
/dev/dsk/c1t4d0
/dev/dsk/c1t1d0
/dev/dsk/c5t8d0
/dev/dsk/c4t2d0
.
.
do a pvdisplay -v on the ones on line 1 and 3
because these are the ones causing the errors

vgdisplay -v vg01

and

ioscan -fnk

just to have a look to make sure everything is actually all okoy.
It could just simply be a disk is slow to spin up.

Bill

[Uday_S_Ankolekar]

I changed I/O time out settings for all the disks on vg01.

pvchange -t 90 device name.
This helped a little bit. The errors in syslog
still displays the same but not so frequent it used to.

dmesg still gives the same error.

Vgdisplay does not show any errors.