System Administration
cancel
Showing results for 
Search instead for 
Did you mean: 

Frequent AdvFS I/O errors and disk failures in HSZ80

Frequent AdvFS I/O errors and disk failures in HSZ80

OS and Hardware Information :

Tru64 Unix Version 5.1B on an ES40 connected to Multi-bus Failover configuration of HSZ80s.

Patching Information :

Patches installed on the system came from following software kits:
----------------------------------------------- - T64KIT0024267-V51BB25-E-20041122 OSF540
- T64V51BB25AS0004-20040616 OSF540
- T64V51BB26AS0005-20050502 OSF540
-----------------------------------------------

Problem :

I have been experiencing frequent HDD failures on the HSZ80 ( my OS also being in the HSZ80 ).

One interesting pattern observed has been the refurbished 36 GB Drives failing mostly ( 5 out of the 6 HDDs failed within the last 1 month were these ones ).

Recently i am also begining to see AdvFS I/O errors on the /var/adm/messages file as in the attachment.

As suggested in the messages file, i have tried increasing the setting of AdvfsIORetryControl from 0 to 5.

The HSZ80 UNIT seen as dskx to which this domain on which the AdvFS I/O error occurs is a RAID 0+1 Storageset.After having obtained the name of the file on which the AdvFS I/O error occurred by using tag2name, i would like to know if it is possible to find the physical hard-disk on which this file resides.

Also, i would like to kno what is the optimal value of AdvfsIORetryControl ?.

Any suggestions / thoughts to close this chapter once and for all, please pour in.

Thanks in advance.

Srivathsan A

3 REPLIES
Nikki-8)
Occasional Contributor

Re: Frequent AdvFS I/O errors and disk failures in HSZ80

Are the HDD failures on the same slots each time?

Re: Frequent AdvFS I/O errors and disk failures in HSZ80

Hi Nikki,

Unfortunately not :-(

Thanks
Michael Schulte zur Sur
Honored Contributor

Re: Frequent AdvFS I/O errors and disk failures in HSZ80

Hi,

from the log you can see that it is
/dev/disk/dsk6a
I haven't worked with the HSZ80.
If it has a console port, connect a terminal and watch for error messages.
How old are the disks?
Are they from roughly the same production batch?

greetings,

Michael