who can tell me which can cause the different of the two command ?

[xxl_1]

My two machine model is ia64 hp server rx5670 , the Os is HP-UX B.11.23 ,

I send out the command "# dmesg |more "

the output is :

Feb 12 12:28
...
msgcnt 27 vxfs: mesg 001: vx_nospace - /dev/arch1/lvol1 file system full (1 bloc
k extent)
msgcnt 29 vxfs: mesg 001: vx_nospace - /dev/arch1/lvol1 file system full (1 bloc
k extent)
msgcnt 31 vxfs: mesg 001: vx_nospace - /dev/arch1/lvol1 file system full (1 bloc
k extent)
......

but I send out the command " # bdf "

the output is :

/dev/arch1/lvol1 52416512 11615916 38250787 23% /oradata/crm/arch1


who can tell me which can cause the different of the two command ?

[Reshma Malusare]

Hi,
Is there any time diference between 2 commands.
please,use command bdf with .
date;bdf

& dmesg

[Reshma Malusare]

Hi,

dmesg -- collect system diagnostic messages to form error log.

dmesg looks in a system buffer for recently printed diagnostic messages and prints them on the standard output. The messages are those printed by the system when unusual events occur (such as when system tables overflow or the system crashes). If the - argument is specified, dmesg computes (incrementally) the new messages since the last time it was run and places these on the standard output.

The system error message buffer is of small, finite size. dmesg is run only every few minutes, so there is no guarantee that all error messages will be logged.

[Patrick Wallek · Accepted solution]

The dmesg command retrieves data that is currently in the dmesg log buffer. However, that data may not necessarily be data that is applicable at the time you run the command.

dmesg stores NO time/date information, so you can really never know EXACTLY when the messages in dmesg occurred.

It is entirely possible that the file system full message you see is hours or maybe even days old.

A better place to check is /var/adm/syslog/syslog.log. That will show a time/date stamd and you WILL know exactly when an error occurred.

Try this to find out when your file system full actually occurred:

# grep "file system full" /var/adm/syslog/syslog.log

[A. Clay Stephenson]

Here's a question for you. Suppose that the filesystem (pick one, it doesn't matter which) used to log error messages fills up? Where are you going to write the messages that this filesystem filled up? That is the purpose of the circular dmesg buffer. It's in memory so it can always be written to but the size is finite. In any event, the filesystem filled up; the offending process unlinked the file and terminated and the filesystem space was returned.

You should have filesystem watchdog daemons in place to inform you when a filesystems reaches a certain threshhold. You can then take action BEFORE there is a problem.

[Reshma Malusare]

Hi,
Please assign points if your query is solved.

Thanks & Regards
Reshma

[James George_1]

Hi

Your dmesg is referring some old File system full message.
your bdf is the correct / present one .

Rgds / James

[Bill Hassell]

It is imperative that dmesg be logged with timestamps. This is done with the - option, something like this in cron:

* 0,15,30,45 * * * dmesg - >> /var/adm/dmesg.log

Now, all new dmesg messages will be logged with the time of day. The - option records nothing if there are no new messages and adds the time stamp when there are new messages. man dmesg

[xxl_1]

but the "Feb 12 12:28 " in the output of
command "# dmesg |more " is the current time ?

how to explain it ?

[Raj D.]

xxl,

You can also refer the syslog.log for filesystem full error and collect the timestamp and compare with the current bdf, and current time.

It's always a good idea to use some monitoring tool or script that will notify you whenever filesystem reach 90% or more.
You can get plenty if monitoring scripts from itrc.


Cheers,
Raj.