disk and file system integrity check

[Vaish Raj]

Hi,

1)Is there any need to run "fsck" regularly on filesystem to make sure filesystem is not corrupted and to avoid future problems ?

For some reason or other we are rebooting our development server regularly(once a month). I believe HP-UX dies "fsck" check on all filesystems during booting. Is this enough to avoid potential file/disk problems.

2) Are there any other procedures/checks to be done regularly to avoid potential file/disk problems.

Thanks
Raj

[Bill Hassell]

1. There is no reason to run fsck regularly if you keep your system up to date on patches. There is no resaon to reboot HP-UX systems except during scheduled patch installations. The reason someone decided to do this is out of habit from other Unix server that had lots of problems that were not being fixed. Filesystem corruption is almost unknown in HP-UX except due to outside influcences (ie, root user makes a mistake).

2. The MOST important thing to do is to mirror ALL of your disks. Disk failures will occur due to hardware problems...that has never changed although the frequency is much less than a decade or two ago.

The other part for protecting your disks and files is to severely limit the number of people with the root password and do not allow untrained administrators or consultants to have the root password.

2.

[Sivakumar TS]

Hi Raj,

As long as there is no improper shutdown or improper unmounting there is no need to run fsck.

May be to maintain optimal performance on busy file systems,

To defragment a JFS file system using fsadm, execute the following to perform both directory and extent reorganization and to generate reports before and after reorganization:

fsadm -d -D -e -E /mount_point

Regards,

Siva.

[Sivakumar TS]

Hi,

you can also schedule a regular check by adding a cron job.

For example, to defragment every evening at 9 p.m. all the extents and directories within the file system mounted at /home, include the following entry in a file used by cron

0 21 * * * fsadm -d -e /home

Regards,

Siva.

[Yogeeraj_1]

hi Raj,

i think it will be wiser to spend more time on the following:
- Redundancy
- Backup and test Recovery

Frequent reboots hinder uptime.

kind regards
yogeeraj

[Mridul Shrivastava]

If fs type is vxfs it checks only log replay so it won't take much time but if it is hfs fsck takes long time as it checks whole FS. Anyway fsck is not required everytime.

[AshishJain_USA]

As suggested by most of other posts, redundancy ( mirroring ), backups and recovery testing are the best ways to avoid any unforeseen from happening.

Other good options in a mirrored environment would be do write scripts that would look for any discrepancies between "Cur PV " and "Act PV ". You may also look at lvdisplay outputs to ensure there are no " stale " extents.

rgds....

[Vaish Raj]

Hi Ashish,

Waht type of scripts you are looking at to check mirror disks. Can you upload them.

Thanks

[Alzhy]

Vaish,

Peace and Clean Air!

Normally, you do not need to run fsck or any other filesystem integrity checks. This is because ('cept ofr /stand on 11.11), all filesystems are "journalled" via JFS (a.k.a. VxFs).

If you mean maintenance like defragmentation - it depends on the filesystem use. For fileshare use and filesystems with very dynamic content (lots of deletes/creates,etc - like a dev/build environment) - you may occasionally run a defrag via fsadm as a previous post suggested.

But as far as regular fsck's (via reboots0 -- no need.

IMHO.