Log Backup with DeviceType pipe
i read the tutorial for backup with pipe:
-Log Backup to a Pipe
-Defining a Backup Template for Log Backup to a Pipe
I become the following error on shell in dbmcli:
dbmcli on N4S>backup_start DemoPipeLog
-24988,ERR_SQL: Sql error
-903,Host file I/O error,No more devices
17,Servertask Info: because Error in backup task occured
10,Job 1 (Backup / Restore Medium Task) [executing] WaitingT137 Result=3700
6,Error in backup task occured, Errorcode 3700 "hostfile_error"
On DBMGUI in Backup-History the following error:
Error: No more devices(-903)
I think I understand the "pipe" strategy.
Can someone help me to explain this "pipe" strategy.
Markus Doehr replied
A pipe is a special device file on the operating system.
If you backup to a pipe, you need a program "on the other side of the pipe", that picks the data. I use that mechanism to do system copies:
- create one (or more) pipes on operating system level (on source and target)
mknod /tmp/pipe1 p
- create a backup medium, that writes to the pipe and reads from the pipe on the other system
- start the backup
- take the data out of the pipe and put it on a second pipe on a remote system
dd if=/tmp/pipe1 | rsh <remotesystem> dd of=/tmp/pipe1
- restore the backup on the second system using that pipe
If you are only local you could do
dd if=/tmp/pipe of=<some-file-on-the-filesystem>
There are backup tools (HP Dataprotector, EMC/Legato Networker) that use that mechanism to do backups.