Menu
▾
▴
[Bacula-bugs] [bacula 0000689]: 2drive-incremental-2tape failure
|
From: <bac...@li...> - 2006-10-11 07:47:09
|
A BUGNOTE has been added to this bug. ====================================================================== http://bugs.bacula.org/bug_view_advanced_page.php?bug_id=0000689 ====================================================================== Reported By: Michael Brennen Assigned To: ====================================================================== Project: bacula Bug ID: 689 Category: Storage Daemon Reproducibility: always Severity: major Priority: normal Status: feedback ====================================================================== Date Submitted: 10-06-2006 16:41 PDT Last Modified: 10-11-2006 00:46 PDT ====================================================================== Summary: 2drive-incremental-2tape failure Description: This is a failure of the 2drive-incremental-2tape regress test on a 2 drive Q47. As best I recall, and it has been a couple of weeks since I saw this on the production system, this is the bug I was seeing, the 3903 failure. Glad to supply more info as needed. ====================================================================== ---------------------------------------------------------------------- kern - 10-07-2006 03:47 PDT ---------------------------------------------------------------------- Please reupload all your attachments adding .txt as the extension, otherwise they are cannot be read by Mantis. Please do not use underscores in the names as was the case in another bug. Your instructions are incomplete. They tell me to run the 2drive-incremental-2tape test, which works fine here. Please refine your "Steps to Reproduce", or explain why the test is failing (perhaps this will be more obvious when I see your uploads, but at least it would be helpful to point out in which file I should look). ---------------------------------------------------------------------- Michael Brennen - 10-07-2006 10:29 PDT ---------------------------------------------------------------------- Unfortunately I don't have the original log files any more. I completely forgot about the .txt extension requirement. I am rerunning the 2drive-incremental-2tape now. ---------------------------------------------------------------------- Michael Brennen - 10-07-2006 11:46 PDT ---------------------------------------------------------------------- I was able to download the files and reupload them with new names. Hopefully the renamed files will be visible. The log1.txt file has this error; this is the error I was seeing a couple of weeks ago. 06-Oct 17:48 localhost-sd: NightlySave.2006-10-06_17.42.51 Fatal error: 3992 Bad autochanger "load slot 2, drive 0": ERR=Child died from signal 15: Termination. 06-Oct 17:48 localhost-fd: NightlySave.2006-10-06_17.42.51 Fatal error: job.c:1732 Bad response to Append Data command. Wanted 3000 OK data, got 3903 Error append data This morning I reran the 2drive-incremental-2tape test twice and it did not fail. That is a bit frustrating, as it did consistenly fail a couple of days ago. I am running current stock centos4 kernels, so I do not look for the sorts of OS vagaries that you were seeing with the Suse kernel. I will continue to work with it and keep you posted. ---------------------------------------------------------------------- Michael Brennen - 10-07-2006 14:37 PDT ---------------------------------------------------------------------- A second instance of failure of 2drive-incremental in log1second.txt and log2second.txt. Sorry, I just awoke from a nap and groggily forgot the no underscore rule. The library ended up with TestVolume001 in slot one in the changer magazine and TestVolume002 in drive 1; drive 0 was empty. log1second.txt gives the same 3903 append error. Again, I do not know what to give you in additional information regarding the failure and how to reproduce it. If you can ask more questions or point me to different areas I will help as I can. ---------------------------------------------------------------------- kern - 10-08-2006 01:37 PDT ---------------------------------------------------------------------- I am still unable to read via Mantis your uploads. Perhaps you are running some strange code page that is not UTF-8, which might confuse Mantis. In thinking about your problem, I would suggest: - don't spend any more time on the simulator. If it doesn't work it is either because you have a timing problem with your autochanger (my best guess) or the simulator is not simulating something correctly. - If I remember right your autochanger was killed by a signal 15, which indicates to me that either the default timeout for the autochanger or the value you set is too small in some cases. I'd recommend that you double the timeout and re-run the tests. - You might want to pull the latest CVS, I've added code that will show the full output from mtx in the Bacula log when there is an error. This may show something. However, in your case, if it is getting killed by a timeout (i.e. Bacula is forcing the script to die), you won't get much more info. Sorry, I don't have time today for more details. ---------------------------------------------------------------------- Michael Brennen - 10-10-2006 19:28 PDT ---------------------------------------------------------------------- I'm not sure what to think about the txt log file uploads. I am running a standard Firefox on a current OSX laptop at home, where I know some of the files were uploaded. I can read back my uploaded files fine. Perhaps the Mac CR only line ending is causing problems???? The log files were created on a linux system. I think I have uploaded files from both my Linux desktop at the office and the OSX laptop. ---------------------------------------------------------------------- ArnoL - 10-11-2006 00:46 PDT ---------------------------------------------------------------------- Just another report concerning file readability: log2second.txt does not show inside the browser, but once opened in an editor it contains ASCII text (reporting a failed regression test restore). This is with Firefox on Windows XP. Bug History Date Modified Username Field Change ====================================================================== 10-06-06 16:41 Michael BrennenNew Bug 10-06-06 16:41 Michael BrennenFile Added: bconcmds 10-06-06 16:41 Michael BrennenFile Added: log1.out 10-06-06 16:42 Michael BrennenFile Added: log2.out 10-06-06 16:42 Michael BrennenFile Added: q47.conf 10-07-06 03:47 kern Bugnote Added: 0001927 10-07-06 03:47 kern Status new => feedback 10-07-06 10:29 Michael BrennenBugnote Added: 0001928 10-07-06 11:37 Michael BrennenFile Added: bconcmds.txt 10-07-06 11:37 Michael BrennenFile Added: log1.txt 10-07-06 11:38 Michael BrennenFile Added: log2.txt 10-07-06 11:38 Michael BrennenFile Added: q47.txt 10-07-06 11:46 Michael BrennenBugnote Added: 0001930 10-07-06 14:29 Michael BrennenFile Added: log1_2.txt 10-07-06 14:29 Michael BrennenFile Added: log2_2.txt 10-07-06 14:37 Michael BrennenBugnote Added: 0001932 10-07-06 14:37 Michael BrennenFile Added: log1second.txt 10-07-06 14:38 Michael BrennenFile Added: log2second.txt 10-08-06 01:37 kern Bugnote Added: 0001934 10-10-06 19:28 Michael BrennenBugnote Added: 0001937 10-11-06 00:46 ArnoL Bugnote Added: 0001938 ====================================================================== |