/var full

[watermelonyu]

This is HP Tru64 5.1A cluster. If one node's file system '/var' full, it will cause one node crash?

[watermelonyu]

Hello:



I nener heard about this situation, but if you found your /var is near full, you should keep good

admin practice to find out why /var is near full, is there any abnormal process writing data, or there is some system abnormal cases because this is the major system log file system, such as /var/adm/binary.errlog and /var/adm/syslog.dated/ dir.



Best Regards

Danny Lin

[watermelonyu]

Some scheduled jobs produced many root's mail suddently. This caused the root's mail box full and /var full. The file system full '/var' caused the server crashed? Have any comment?



Thanks



Eric

[watermelonyu]

Hello:



What are your scheduled jobs? Are they run

from cron or others? And could you please check

/var/adm/crash/crash-data.x for your crash info,

and provide it for us to check?



Best Regards

Danny Lin

[watermelonyu]

/var 應該是share cluster file system

如果滿了應該也是兩個nodes都會影響才對阿

請問是哪一個node crash

當時/var 的owner 是哪一個node

[watermelonyu]

Hello:



我也是認為crash不單純是/var full, 所以請post

(使用附檔方式)/var/adm/crash/crash-data.x(最近

一版), 以便判斷!



Best Regards

Danny

[watermelonyu]

It crashed on 9/10.

However, we found nothing under /var/adm/crash.

Pls see below,

-----------

#pwd

/var/adm/crash

#ls -l

total 1

-rw-r--r-- 1 root system 2 Sep 7 2003 bounds

------------



Thanks

[watermelonyu]

Hello:



No crash files are generated, please check

/var/adm/messages, or post it as attachment.



Danny

[watermelonyu]

It crashed at 14:30 9/10.

[watermelonyu]

Hello Eric:



請參考附檔: 說明如下

1. hostname has been changed to protect you!

2. /var is full from Sep 10 03:26:28 so when

it crashed, there is no space to write crash

data to /var/adm/crash, so we can't perform

more analysis.

3. There are one Hard disk error after reboot,

please call HP 0800211666 for hotline service,

or call your account engineer to check Hard

disk.

4. I think your /var is full because root got

so many mails that may be related to /var is

full or other events, so keep in mind to check

/var space so if it UNFORTUNATELY CRASHED AGAIN,

we have footprint of crash for analysis.

5. If you got crash UNFORTUNATELY, the right

rule is to call HP hotline, not post it in 討論

園地.



Danny

[watermelonyu]

Hi Eric,



There are 2 actions need to apply by yourself.

1) Please solving the full FS first on /var or extending its size. Due to no more free spaces for recording all of system logs even the crash dump file into /var/adm/crash while it encountered a crash dump at 14:30 9/10. So that HP

will not be detailed for further analysis.



2) Please upgrading the latest OS patch kits to V5.1A/PK6. Because more bugs within V5.1A/PK3 or PK4 when it got a

panic with KMF (kernel memory fault) as your system.



Best regards,

Richard.