Compressing oracle archive logs

[unixdaddy]

We have a production server on which we compress the archive logs in order to save space, we use the compress command.

We then built a test server from a backup of the production machine. Once this was done we ftp'd the redo logs from the production machine over to the test machine, uncompressed them and then attempted to roll the database forward. However the database kept saying that the archive logs were corrupt.

Is the problem with oracle and the way it is writting the logs out or is it the fact that the logs were compressed on a different machine?

[Praveen Bezawada]

Hi
Compressing on a different machine would not have affected the archive logs. Have you ftped in binary mode?
This looks like an oracle problem.

...BPK...

[unixdaddy]

Yep, we FTP'd it in binary mode.

[harry d brown jr]

run a cksum on your "production" servers archive log, then compress it, ftp it to your test system, then on your test system do another cksum to see if the file changed. If not, then it is an oracle thing.

[Stephen Serbe]

I looked into compressing Oracle's archive logs a year or so back... Oracle would not support the compress/zip process... Check with them to see if their stance has changed. Good luck.

[Don Spare]

I was concerned about the implications of this thread so I went looking in the Oracle Metalink forums and found that Oracle does support compressed archived logs. They suggested that if the logs are corrupt after uncompressing them that the problem may be that they were corrupt to begin with or perhaps the disk they ended up on had a problem.

Hope this helps.

If you have access to the Oracle Metalink please search for 'compress archive log' and it will return a list of articles/threads that discuss this in greater detail.

[unixdaddy]

I'd like to thank you all for your contributions, I've only just awarded points because I wanted to see if there were anymore contributions, thanks. Our Oracle guys are looking into it at the moment.

P.S

Has anyone noticed that there are alot of threads where no points have been allocated. I would encourage those who ask questions to give points!!!

[Alan Riggs]

We gzip oracle logs. The only time I have seen a problem occurred when the automated process started zipping a file as it was being generated. This resulted in a corrupted zipped file *and* a second, incomplete, uncorrupted archive log. Ugly.

My zip routine now excludes the most recently generated archive log from teh process and I have had no problems.

[unixdaddy]

Thanks. The script we use excludes the most recent logfile from the compression routine, so that's OK.

[Thierry Poels_1]

hi,
we zip our archived logs too, never had any problems with it.
But instead of skipping the last file, we get the list of files, wait a few seconds (should be enough to have the last one written completeley), and then zip the list.
regards,
Thierry.