Sorry for the long Delay.
There seem to be some very strange and not reproduceable situations
where a specific combination of dump-files in the holdingare, Moonphase and such things lead to the effect you see.
I had this happen on two different configs, on two different machines a few month/years ago.
What was the solution to it?
I moved one dump-image out of one of the folders in the holdingdisk, started amflush, and it just did what it was supposed to do.
After it finished i moved the image back into the correct holding dir, did another amflush and you wouldn't beleave it, it was flushed without any problems.
So you have two pathes to go:
1.) get out your debugger and dig trough the processes amflush spawned
to see where it is stuck and fix that strange bug,
2.) move one of your holdingfiles awy, and flush your backups to tape,
leaving the bug for another one to fix...... ;-)
Christoph
Jon LaBadie schrieb:
On Fri, Dec 10, 2004 at 05:47:22PM -0000, James Marcinek wrote:
Does anyone have experience dealing with issues regarding the amflush not working properly? The system in a RH9 box. I know there's more recent versions than what's being used but I cannot find the later rpm's that are necessary. I can provide what ever information is necessary to help:
The amflush seems to initiate but all the children processes sit idle and do not dump to tape (did amverify afterwards...)?
I follow the instructions and read the man page for amflush. I haven't found much subject matter regarding my issue on the web.
Does amstatus tell you anything interesting?
Have you looked through the debug files, probably in /tmp/amanda, for any indication of a problem? (It might be good to rename the or empty the current /tmp/amanda and rerun amflush. Otherwise there could be tons of irrelevant files)
Is there anything to "amcleanup"?
Do your system log files indicate any problems with accessing the holding disk(s) or tape?
