Backing information

[Rob_4]

I have attempted using tar for a backup of files but always receive an error 5, looking at the backup listing it shows acl info not being backed up. So I tried fbackup only to receive error 4 warnings that the log files are open and aren't backed up. I need a solution that backs up all data and does not produce an error code from cron. Suggestions?

[Craig Rants]

What O/S version are you using? If you are using 11.11 then ignite would be an option. Before that however you can only do vg00. Also when are you seeing an error 5? Is that a return code 5? I did see a return code of 5 in the tar man page so I am a little confused on that one.

C

[Sanjay_6]

Hi Rob,

Error 5 from tar is a non-fatal error,

http://us-support.external.hp.com/cki/bin/doc.pl/sid=050ef6b60caed40a61/screen=ckiDisplayDocument?docId=200000038250707

Here is the detail on fbackup error codes,

http://us-support.external.hp.com/cki/bin/doc.pl/sid=951155e30fee7983e3/screen=ckiDisplayDocument?docId=200000047385470

hope this helps.

Regds

[harry d brown jr]

Using fbackup, you can specify the number of times to retry on an active file before skipping it.

Also, it's a log file, and if it is fbackup's so who cares?

Another way is to mirror everything, split it, mount it, backit up, then remirror it.

Furthermore, you can purchase commercial products that will do the same.

live free or die
harry

[James R. Ferguson]

Hi Rob:

'fbackup' is designed to retry files that are changing while a backup is proceeding. It does this by noting the timestamp of a file as it starts to copy it to tape. When a copy has been transferred, 'fbackup' compares the timestamp on the tape to that on the disk, and if different, marks the tape copy "bad" and proceeds to retry the process 'maxretries' times. The default for 'maxretries' is five. See the man pages of 'fbackup' for more details.

In any event, retries are counted as warnings by 'fbackup' and result in a return code of four (4).

I suspect that you are using SAM to do your backups. If you are, I believe that SAM patch PHCO_23876 will obviate the problem. This patch does not require a reboot to install.

Regards!

...JRF...

[Rob_4]

Sorry folks, I'm using HP 10.20. Also point 1, the logging is for a financial app that runs 24x7, so they are important at least as a snapshot daily. Piont 2, The backups are run from cron and therefore the rc codes are a nagging indication of an error condition. I guess what I really need is without going to sungle user or doing a shutdown of the app what can I use to ensure complete backups and limit warnings.

[A. Clay Stephenson · Accepted solution]

Okay Rob, how about a plan B. Use fbackup (this will solve your ACL warnings) but before backing up (as part of your cronjob) does an OnlineJFS snapshot mount. This will take only seconds and you will have a read-only (i.e) no logs changing version of the data.

[harry d brown jr]

Have you thought about using mirrors and splitting them for backups, which requires no "down" time to split?

live free or die
harry

[Frank Slootweg]

It is important to know that open/busy (fbackup's term "active") files *are* backed up by fbackup. It just 'complains' about them, so you know they were active and probably (partially) corrupt when restored. As far as I know these active-but-still-backed-up files can also be *restored*, but it would be wise to confirm that 'theory' by doing a trial restore. With the frecover option -F or -X you can restore them to a different location.

As to "error 4 warnings" (assuming RETURN VALUE 4), you say that the backups are run from cron, but that does not say whether or not this is a SAM scheduled backup. So who/what made the crontab entry, 'you' or SAM? If SAM, then you need the SAM patch which James refered to, so that *warnings* will not be considered to be *failures*.