BillIrwin Posted January 11, 2005 Report Share Posted January 11, 2005 I'm running under XP Pro SP2 with Retro MultiServer 6.5.350. My backups always contain -1020 sharing violation errors on the same files in these two profiles under C:\Documents and Settings. I've tried to exclude them every way possible in the Selector, but the errors always show in the log. Since these are system profiles I don't care to back them up and would like to have a backup with no errors, if that's possible. Is there a magic method of getting Retro to ignore these two profiles? Bill Link to comment Share on other sites More sharing options...
natew Posted January 12, 2005 Report Share Posted January 12, 2005 Hi What edition of Retrospect are you using? Do you have the open file backup option? Thanks Nate Link to comment Share on other sites More sharing options...
BillIrwin Posted January 12, 2005 Author Report Share Posted January 12, 2005 Using 6.5.350 trial on WP Pro with NTFS filesystem. I've turned off Open File Backup because this generated an error about the option not being available. I assumed it's only available on the licensed version, but I'd like to test this as well. I'll turn it back on and see what happens on the next backup. Thanks. Bill Link to comment Share on other sites More sharing options...
BillIrwin Posted January 13, 2005 Author Report Share Posted January 13, 2005 With Open File option on, running on Windows 2003 server, backing up Volumes on a client running NT4 server, I get error -1012 (feature unsupported). I get this on each of the two client volumes. Then I get several -1020 (sharing violation) errors. However, there are no sharing errors on the local C: drive, so Open File backup seems to work locally, but not on a client (at least this NT 4 server client running 6.5.136. Bill Link to comment Share on other sites More sharing options...
natew Posted January 14, 2005 Report Share Posted January 14, 2005 Hi What is the format of the disks on the NT client? How are the disks attached to the machine? All features are enabled on the trial version of Retrospect. Nate Link to comment Share on other sites More sharing options...
BillIrwin Posted January 14, 2005 Author Report Share Posted January 14, 2005 The Volumes are the C: and D: partitions (only two) of the NTFS primary disk, which is mirrored to a second disk using NT4 mirroring. These two drives are the Master and Slave on the primary IDE channel. Link to comment Share on other sites More sharing options...
natew Posted January 14, 2005 Report Share Posted January 14, 2005 Hi Interesting. Theoretically this should work. Open file backup supports NTFS volumes on IDE and SCSI busses. I wonder if the mirror is causing problems... I would uninstall the client software and reboot the client machine. Then reinstall, reboot and make sure the roff.sys file exists at c:\winnt\drivers\ If not do a search and see if it is anywhere on the disk. Thanks Nate Link to comment Share on other sites More sharing options...
BillIrwin Posted January 14, 2005 Author Report Share Posted January 14, 2005 Done. The roff.sys file is in C:\winnt\system32\drivers. We'll see what happens tonight. This NT4 server also has the Retrospect 5.5 installed that I was using there until I put 6.5 on the 2003 server. It's in D:\Program Files and the client is in C:\Program Files. Bill Link to comment Share on other sites More sharing options...
BillIrwin Posted January 14, 2005 Author Report Share Posted January 14, 2005 Now it looks like Open File backup is trying. I got two e-mails, one for each Volume, with "Can't use Open File Backup option for Users (D:) on nsaserver, error -3047 (disk inactivity threshold not met)". I got an identical one for C: on the client. This error comes before the Copying line in the log, so it looks like it tries to get inactivity on the disk, fails, then does a backup without Open File. I say this because all the normal sharing violation files are listed. Mind you, some of these are the software router cache files (this server is multi-homed and acts as the Internet gateway). There is always activity. The other are some WINS files and some UPSMON files. All of these functions are continuous. If the successful use of Open File backup depends on getting 5 sec. of inactivity I don't think that's going to happen on this server. What about setting this threshold lower? Bill Link to comment Share on other sites More sharing options...
BillIrwin Posted January 15, 2005 Author Report Share Posted January 15, 2005 Yesterday I turned off "Protect multi-volume datasets" and left the threshold at 5000. Today I got one e-mail about partition C: not meeting the threshold, but no problem with D:. Now I'll try turning down the threshold to 4000 to see if I can get C:. It would sure be nice to have the option, by backup script, of sending an e-mail if there are no errors ... optionally including the log for that run. It bothers me not to hear anything if everything is going OK. If the Launcher service has failed then there wouldn't be any backups and no e-mail, I suspect. Also, I had the e-mail incorrectly configured to authenticate on the SMTP server. This prevented a message from being delivered and there was no log entry to indicate the SMTP failure. Bill Link to comment Share on other sites More sharing options...
natew Posted January 17, 2005 Report Share Posted January 17, 2005 Hi 5 seconds is the general rule for a disk activity timeout. This figure is designed for database backup where even a small amount of corruption in a file can cause serious problems. On a machine with no critical/huge databases you are probably safe going all the way down to 1 second disk write timeouts. A portion of a file may be damaged on restore but it likely won't be. Besides a file that was open at the time of backup will have been changed and marked for backup again next time around. General rule - the higher the threshold the better but cutting it down won't hurt too much. You can configure external scripts to notify you of successful backups. Check the external scripts section of the manual. Nate Link to comment Share on other sites More sharing options...
BillIrwin Posted January 23, 2005 Author Report Share Posted January 23, 2005 Well, I had to crank the threshold down to one second, but today I got my first zero error backup on both remote server volumes and the backup server C:. Thanks for the help getting to this point. Once a normal backup isn't generating any errors, a single error becomes reason to investigate. However, based on the e-mail experience I've had so far, I don't think I'll get mailed if there's a non-critical error. I get e-mail when media is needed and to warn that the open file threshold wasn't met. Bill Link to comment Share on other sites More sharing options...
natew Posted January 27, 2005 Report Share Posted January 27, 2005 Hi You are correct about the email. It only gets sent for critical errors and media notification. FWIW the disk write timeout is designed to make sure any data in the disk cache is included in the backup. If you have cache disabled on your server disks you can keep the timeout low with no problem. Nate Link to comment Share on other sites More sharing options...
Recommended Posts
Archived
This topic is now archived and is closed to further replies.