Start a Conversation

Unsolved

This post is more than 5 years old

1047

April 2nd, 2008 08:00

Time Finder Clones and deactivate

Currently I have a Progress database running on AIX 5.2. When the database is down I run the following SE 6.42 commands to clone it to another AIX host.

symclone -sid 1303 create -file file1
symclone -sid 1303 activate -file file1
on second the target AIX host -
importvg -y b_db hdiskpower2
mount /prod/b/db

file1 is made of the following source and target devices -
101 6DD
227 6ED
237 6FD


The clone session uses the default copy on access. If 5% of the data is accessed on the target side then some of the tracks will be copied from source to target.

What happens when I run the terminate command "symclone -sid 1303 terminate
-file file1" a short time later? I can still access the data on the target side even though only 5% of it was copied (coa) from the source. When I access the other 95% on the target is it still going back to the source devices to get the data even though I did a terminate?

I'm not sure how it continues to access the un-copied data on the target after I do the terminate.

2 Intern

 • 

20.4K Posts

April 2nd, 2008 09:00

are you sure you tested every block on the disk ? :)

from TimeFinder manual:

Terminating a session while the device pairs are in the CopyOnAccess or CopyInProg
state will cause the session to end. If the application has not finished accessing all of
the data, the target copy is not a full copy.

2 Intern

 • 

2.8K Posts

April 3rd, 2008 01:00

Did you unmount and remount the filesystem after terminating clone session ??

If you don't unmount/remount the devices, your "other" AIX host will keep a copy of FS description in its cache.

Try unmounting the FS and fscking it ;-)

-s-

207 Posts

April 3rd, 2008 16:00

Hmmmm still a little strange. I unmounted the fs and ran fsck. I remounted, and it seems that I can still access all of the 135GB.

It seems like somehow it copied all of the data over. This is strange because I did a create and activate with no special paramaters so it shoud have used coa and only copied the small amount that I accessed.

This one may remain a mystery.

2 Intern

 • 

2.8K Posts

April 4th, 2008 01:00

135 gb is the size of the database or of the disks ?? if you issue a df -k how full is the filesystem ??

Try another test .. keep the FS mounted (you already unmounted/remounted it) and find a file that wasn't modified .. I think you can identify them with a "find -atime" to find older (and untouched) files. Verify that the same file is still untouched on the production host (same date/time stamp as in clone). Compare them and see if they are really the same ... I hope they will differ :D

What did you do on both hosts before terminating the session ??

Maybe you have a full legal description of the filesystem on your clones becouse you issued a recursive find that "touched" all inodes, without touching the content of the files, thus copying only a few blocks on the clones, but the "right" blocks to give you the perception of a complete filesystem :D

207 Posts

April 4th, 2008 08:00

ok, I think I got this figured out now. Maybe I messed something up the first time.

This time I started from scratch - I removed the AIX volume groups from the target. I then created and activated the clone from scratch. After imporvg and mount of filesystems on the target I did a cd to the file systems directory. I did an ls command to list out the directory. I could see the 134 database extents that I expected.

I opened one of the extents (with the more command) and looked at it. I then terminated the clone session. I then opened the same first DB extent to verify I could still access it - looks good. I then tried to open several of the other DB extents. I could not open them. This is good. The terminate is working as expected.

Thanks
No Events found!

Top