Space used by snapshot filesystem

[Sukant Naik]

Hi,

We are using Online JFS on a HP-UX system.

I have got a filesystem /user1 of size 4GB. I created a snapshot of /user1 filesystem called /snap1 of size 400MB.

After sometime, the /snap1 system becomes inaccessable because all blocks get used up.

I have got few doubts.
1. I can increase the size of the /snap1 filesystem, but before that is there any way we can find out how much of the /snap1's logical volume is being used.

2. So, if we can detect this, can we dynamically increase the size of /snap1.

Regards,

Sukant

[Bill McNAMARA_1]

Stephen Steels response in this thread will led you know:

http://forums.itrc.hp.com/cm/QuestionAnswer/1,,0x99af26df904bd61190010090279cd0f9,00.html

Later,
Bill

[Bill McNAMARA_1]

mmm,, maybe not that easily afterall!

Later,
Bill

[Sukant Naik]

Hi Bill,

Thanks for the links which you had sent to me. But even after deleting directories inside the snapped filesystem, the snapshot filesystem is showing no difference.

I am still trying here. Any additional info will be helpful.

Regards,

Sukant

[A. Clay Stephenson]

Df's will not really help you in this case. At best, df's can reveal the net difference between the number of newly allocated blocks and removed blocks in the original filesystem vs. the snapshot. Df's can tell you absolutely nothing about the most common use of snapshot buffer blocks - updates. The FIRST time a given block is updated in the original filesystem, one block of snapshot buffer is used. Df's reveal nothing about this use of the snapshot buffer.

10% snapshot buffer size is a little on the low end for busy filesystems but I have seen very few filesystems that require more that about 20% buffer space as long as the snapshots are limited to a reasonable period of time.

[Sukant Naik]

Hi Bill / Clay,

Thanks for your response. Looks like I need to find some other way of doing this.

I am still trying. I shall update this list when I find something good.

Thanks once again for your help.

- Sukant