From: Kern S. <ke...@si...> - 2006-09-18 16:31:52
|
Hello Richard, On Monday 18 September 2006 17:57, Richard Mortimer wrote: > Hi Kern, > > I see you've been busy over the weekend :-) I've just pulled across > current CVS and will give things a test. Yes, I put a lot of time into it. > I note that a few of your > emails asked questions - I'll try to go through them and answer > where you haven't worked it out for yourself. Thanks. I think I probably answered most of them myself, or rewrote the way the code works. > > A few comments below... > > > -----Original Message----- > > From: bac...@li... > > [mailto:bac...@li...] On Behalf > > Of Kern Sibbald > > Sent: 16 September 2006 10:11 > > To: bac...@li... > > Subject: Re: [Bacula-devel] DVD relabel > > > > On Saturday 16 September 2006 10:11, Scott Simpson wrote: > > > 2006-09-16 CVS HEAD > > > > > > I used this version and applied Richard's patch from message > > > <1158361517.8580.89.camel@duncow>. I still can't use the > > second DVD but it > > > doesn't hang in bconsole anymore. Now I get > > > > Yes, I expected that my changes would seriously disrupt the > > DVD code, but it > > gives us a much cleaner base to work from. I'm pleased to > > see Richard's > > patch, which I will integrate today. Concerning your > > problems: perhaps > > Richard will have time to look at it, if not, I will be > > getting to it in the > > next few days, but first, > I know that Scott has reported success but I will have a look > at the logs to ensure that it is something that you've fixed. > > > I have to write some DVD regression > > scripts so that > > we can be sure that the base DVD writing works and that any > > additional > > changes do not re-break something. After that, I can take a > > look at multiple > > volumes. > Is there any way that I can help here? I can try to enumerate > the test sequences that I run through (the ones that have > tended to break things anyway) if that helps. I've written down the steps that I used to test starting from a basic manual label. The problem now is to write regression scripts that use my dvd-simulator. It is a bit harder than ususal since I need to explicitly check that the correct number of parts were written. I'll get back to doing that pretty soon and should be able to work it out. Where you can be the biggest help is testing with a real DVD. :-) > > > > > > >From what I see below is that when Bacula tried to label the > > DVD, it found it > > was blank, so created the label in the spool file, then > > attempted to read it > > back to check it, but apparently (not enough info to be sure) > > tried to read > > it back from the dvd, or nothing was actually written to the > > spool file. > > This probably won't be hard to fix, but is a just a question > > of time ... > Ok, I just had a quick scan and I think it is the "freespace not > calculated" problem. They key is the free_space=0 error message > > > > End of Volume "DVD13" at 0:0 on device "DVDStorage" > > (/dev/hdc) (part_size=0, > > > free_space=0, free_space_errno=0). > > Scott's SD debug logs didn't go back far enough to see where the last > write took place but that is the error that kicked off his problems > in the last set of logs that were sent to bacula-devel. OK, hopefully I've now invalided the freespace flag in enough places so that the problem is now fixed ... Regards, Kern |