How can i list the chunk names used in a backup?

How can i list the chunk names used in a backup? I am trying to write a script to automate the restore process of ontape backups from a variety of different 'source' databases on my test box which has a lot of disk space. I know i can run ontape -r manually etc, and it will list the chunks from the backup header, however i want to script it such that i can read the chunk names first, then create (touch) those chunk names, and finally perform the restore into those chunks. NB: We use DIRECT_IO so this should be simple. Thanks in advance.

I presume you mean STDIO... STDIO means no prompt, so you get no chance to list the chunks pathnames etc... (I presume you mean a 10.xx or 11.x engine...) simply "turn off" STDIO which will then list the chunks for you... To turn off "STDIO" make sure it is not set in the onconfig, and do not use the "-t STDIO" on the ontape command. You now have a problem because ontape is going to prompt for tape insert etc... and you want this automated.... (I don't believe you can get the chunk info any other way (maybe archecker can look at system tables... but I doubt it) The Ksh way is to create a subshell that answers the prompts for you... ie something like this ----------------------------------------- #!/usr/bin/ksh BACKUP=<backup filename> TAPEDEV=<TapeDEvice> tapecntrl() { cr=0 # start the tape cat <$BACKUP >$TAPEDEV & # tell ontape its loaded echo "y" >>$LOGFILE echo "y" # wait for it wait # rewind the tape cat <$BACKUP >$TAPEDEV & # lets stop echo n } BACKUP=<backup filename> TAPEDEV=/tmp/ontape_pipe # set up pipe to control restore rm -f $TAPEDEV mkfifo -m 660 $TAPEDEV chown informix:informix $TAPEDEV # fire the backup pipe tapecntrl | ontape -r >>$LOGFILE 2>&1 wait ------------------------------- Oh of course the rootdbs in the 'new' onconfig is the set to the 'original' rootdbs path... Tell me how you get on... (And wouldn't it be nice if ontape had an option to extract this info using stdio )

Hi, thanks. Your suggestions gave me an idea; I realised that I can get the header info by setting TAPEDEV to the local ontape backup file, running ontape -r and passing in auto answers, then finally piping the output through grep and lastly dumping into a file. SPACES=/tmp/dbspaces CHUNKPATHS=/tmp/chunkpaths # Find the dbspaces and chunk paths being used in the tape backup ontape -r < /opt/informix/database_restore.inputs | grep ' informix ' | awk {'print $8'} > $SPACES ontape -r < /opt/informix/database_restore.inputs | grep PO-B | awk {'print $7'} > $CHUNKPATHS Contents of 'database_restore.inputs'; y n n This now means I have two files containing purely just a list of the chunks and dbspaces used in the backup file. I have used this list to touch a set of cooked chunks for restoring. The next challenge I have now is I want to have those chunks exist in a directory which is different to that on site. E.g. on the production server where the ontape backup to file was taken; chunks reside in /data/, however on my test box I want to have those chunks exist in /data/customername/. When I try the restore however, ontape moans that ./rootdbs is not the same as /data/rootdbs etc. As dirty work around I wrote my script to create a bunch of temporary symlinks in /data/ pointing to /data/customername/, perform the restore and then drop the symlinks, but this is still not working quite right and the restore assert fails! Is there a special ontape parameter i can pass maybe that says, ignore what you think and just do the restore?

Symbolic links works for me! Historical best practices had all chunks as symbolic links, so you could move chucks. It is obviously still a valid, worthwhile practice. And no, there is no special parameter to make ontape ignore the rootdbs... By the way, I wrote a similar script but used dup_spaces.sh off iiug repository to first generate the chunks from my source database, then I used sed to replace the common lead pathnames to fit my target environment... I had thought of the same idea as you but had onbar in use as well (and onbar can't list the chunks... Now would't it be nice if...) Robert

Hi Robert, Well i got it all working. The failures turned out to be due to a completely unrelated simple issue. I have now got the script to ask which customer you are trying to restore, builds all the correct environment variables etc, asks where your backup is, checks it, updates onconfig with the tape location, reads that database version the backup was from, (Haven't managed to get the correct engine to be auto switched in yet), reads the chunks used, creates them in a customer specific directory, creates the needed symlinks and then finally does the restore. It is almost complete, still a bit rough and ready but it works. I wanted to do this to help speed things up because every time we need to upgrade a customer we have to take a copy and simulate the upgrade. I am a newbie here so i didn't realise there were code samples people have posted. I will go and have a look right now :) Thank you for your thoughts :) Andy