On Saturday 30 September 2006 13:55, Michael Brennen wrote: > On Sat, 30 Sep 2006, Kern Sibbald wrote: > > > On Saturday 30 September 2006 00:26, Michael Brennen wrote: > >> On Friday 29 September 2006 04:36 pm, Kern Sibbald wrote: > >> > >> I am glad you were finally able to resolve the kernel crash; that > >> sort of non-self-induced bug is really frustrating and > >> disfocusing. Just for information so these are not lost, I will > >> note again a couple of cvs bugs that I reported a few days ago. > > > > Yes, thanks. In addition to consuming an enormous amount of time, > > it almost killed me off. When I have a serious bug like that I > > cannot sleep until it is resolved -- unfortunately, I didn't > > recognize it as a kernel problem soon enough -- of course, I > > recognized the kernel was crashing, but I thought that Bacula was > > doing something bizarre, which seems not to be the case. What was > > misleading was that the last beta works fine, the current code > > fails. I suspect it has to do with memory placement > > (bigger/smaller program, I/O that crosses a page boundry, paging > > ...????). > > I fully understand, as do probably many on this list. I have passed > many erratic hours with situations similar to yours. Once a problem > gets to that stage there is not much for it except to drive it to > completion. > > > >> You mentioned a few days ago that I might need to supply some > >> sort of regression script test to demonstrate that. I don't know > >> what such a regression script would look like or where to find > >> examples. If you can point me to a section in the docs I will be > >> glad to supply what information need. > > > > Yes, anything involving multiple drives is so terribly complicated > > that I'm obligated to ask for a test case. > > > > The first thing to do is to pull down the "rescue" module from the > > Bacula CVS. If would be *much* better if you can check it out of > > the CVS since you will be able to easily get any changes I make. > > If that doesn't work, let me know and I'll send you a tar file. > > > > Once you get it, read the README. It should be relatively > > self-explanatory. At that point, you should be able to run the > > regression scripts. Most likely you will see how easy it is to > > write one, if not, let me know and I'll explain some more. > > > > I'd recommend that you submit your multiple drive problem as a bug > > report, and attach a listing as a xxx.txt file with the SD set to > > debug level 50 (this should show us what the reservation system is > > doing). > > I have the CVS of 'regress' pulled down. Since it destroys the > tables I will have to set up a test system for this. I am willing > to do so and have the equipment but it will take a bit.
If you get depkgs from Source Forge then configure regress to use SQLite and point it to the version you built in depkgs. See kern.conf, comment out the mysql WHICHDB and uncomment the sqlite WHICHDB or better yet only include the sqlite WHICHDB in your conf file, then you will be totally independent of any production database. A second level security is to ensure that under your personal userid you have no access privilege to your production db. This is how I do it on one an older Solaris machine where I test. As long as you have a production MySQL and use only a sqlite conf in the regress script, you should never have any problems. Be careful though, that's why the warning is there. > > In the interim, would it be helpful to simply run the SD on the > current system with a debug level of 50 and send the logs? If not > that's okay, I'll set up the test. That may help, but it would be best to submit all info in a bug report attaching the log as xxx.txt (if you don't use .txt as the extension, they cannot be viewed from the bugs program -- Mantis). ------------------------------------------------------------------------- Take Surveys. Earn Cash. Influence the Future of IT Join SourceForge.net's Techsay panel and you'll get the chance to share your opinions on IT & business topics through brief surveys -- and earn cash http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV _______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
