Bacula-users

Re: [Bacula-users] Bacula filling the / (root)

2009-06-01 08:59:17
Subject: Re: [Bacula-users] Bacula filling the / (root)
From: Johan van Vliet <baculaml AT itmon DOT nl>
To: Martin Simmons <martin AT lispworks DOT com>
Date: Mon, 01 Jun 2009 14:54:55 +0200
Martin,

Thanks. I should have thought about these tools myself. But I didn't so 
thank you for that!

It looks like you are on to something.. This time the file descriptor was 
15 and if I look at the 'kdump' output is show;

 15007 bacula-dir GIO   fd 15 wrote 1024 bytes
       "SQLite format 3 
........................................................................... 
...........CREATE TABLE batch (FileIndex integer,J\
        obId integer,Path blob,Name blob,LStat tinyblob,MD5 tinyblob)"

(Replaced \0 and the like with .)

So yes; this is obviously a Sqlite file. Now "all" I have to do is figure 
out what kind of file (Sqlite has several types of temp file) it is and what 
option to move it goes with that type.

Based on the file "cats/sql_create.c" (my_batch_start) in the bacula source 
I would guess it is a "temporary database". 

Thanks! I needed this info.

J.






Martin Simmons writes:

>>>>>> On Sat, 30 May 2009 21:35:06 +0200, Johan van Vliet said:
>> 
>> Hi... 
>> 
>> A couple of weeks ago posted the message below. Today I upgraded the system 
>> to OpenBSD 4.5 (Flashboot) and installed "lsof" inorder to tackle to 
>> problem. 
>> 
>> If I run "lsof" every few minutes with a full backup I see one line that 
>> gets my attention:
>> COMMAND     PID     USER   FD   TYPE     DEVICE  SIZE/OFF     NODE NAME
>> ...
>> bacula-di  1248   bacula   14u  VREG       17,0  30902272     6173 / 
>> (/dev/rd0a)
>> ...
>> 
>> This is exactly the size of the space that I'm "missing" in root but I have 
>> no clue why bacula-dir does this. As I said in my previous mail: the working 
>> directory for bacula(-dir) is: /mnt/archive/working
>> 
>> Any advice on how to trace this futher and avoid this "writing" to root?
> 
> The lsof output (slash followed by the device in parens) means that the file
> was deleted but still open, so perhaps it is a temporary file used by Sqlite?
> 
> Is it always fd 14?  Is it open only when a backup is running?  If so, try
> using ktrace/kdump to see what opens or unlinks it.
> 
> __Martin
> 
> 
>> 
>> J.
>> 
>> > Hi,
>> > 
>> > I have been using bacula/SQLite for a couple of months now (almost a year) 
>> > and it has been running fine up till a week ago. I run a schedule of:
>> > daily incremental
>> > weekly (sunday) differential
>> > monthly full.
>> > 
>> > On March 1st I upgraded my OS (OpenBSD 4.3 build as a flashboot image) to 
>> > OpenBSD 4.4 (also flashboot)
>> > 
>> > Since then my differentials (incrementals are fine) fail. They result in 
>> > an 
>> > SQL error causes by an disk/IO error. When I run dmesg on the bacula 
>> > machine 
>> > apparently the disk was full:
>> > uid 250 on /: file system full
>> > 
>> > However I have configured bacula to use working on /mnt/archive. So it 
>> > should not even try to write in /. The /mnt/archive (I know: bad name) has 
>> > plenty of space (30% in use)
>> > 
>> > What could be writing outside of /mnt/archive/working ? I can only find 
>> > one 
>> > reference to /tmp in the config file and that is for a restore job. Not 
>> > the 
>> > case here.
>> > 
>> > It is possible/likely that my Old 4.3 ramdrive image had a bit more space 
>> > in 
>> > /.
>> > 
>> > Extra info;
>> > Hardware : PC Engine (so no memory upgrade possible)
>> > OS       : Flashboot (mindrot.org) OpenBSD 4.4 (so ramdrive OS)
>> > Bacula   : 2.2.8 (latest OpenBSD)
>> > Database : Sqlite
>> > Storage  : File (USB disk mounted at /mnt/archive)
>> > 
>> > $ df -h
>> > Filesystem     Size    Used   Avail Capacity  Mounted on
>> > /dev/rd0a     19.3M   16.0M    3.3M    83%    /
>> > mfs:18385     14.5M   10.1M    3.6M    73%    /usr/local
>> > /dev/sd0a      458G    131G    305G    30%    /mnt/archive
>> > /dev/wd0a      3.7G   13.5M    3.5G     0%    /flash
>> > 
>> > J.
>> 
>> ------------------------------------------------------------------------------
>> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT 
>> is a gathering of tech-side developers & brand creativity professionals. Meet
>> the minds behind Google Creative Lab, Visual Complexity, Processing, & 
>> iPhoneDevCamp as they present alongside digital heavyweights like Barbarian 
>> Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com 
>> _______________________________________________
>> Bacula-users mailing list
>> Bacula-users AT lists.sourceforge DOT net
>> https://lists.sourceforge.net/lists/listinfo/bacula-users
>> 
> 
> ------------------------------------------------------------------------------
> Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT 
> is a gathering of tech-side developers & brand creativity professionals. Meet
> the minds behind Google Creative Lab, Visual Complexity, Processing, & 
> iPhoneDevCamp as they present alongside digital heavyweights like Barbarian 
> Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com 
> _______________________________________________
> Bacula-users mailing list
> Bacula-users AT lists.sourceforge DOT net
> https://lists.sourceforge.net/lists/listinfo/bacula-users

------------------------------------------------------------------------------
Register Now for Creativity and Technology (CaT), June 3rd, NYC. CaT 
is a gathering of tech-side developers & brand creativity professionals. Meet
the minds behind Google Creative Lab, Visual Complexity, Processing, & 
iPhoneDevCamp as they present alongside digital heavyweights like Barbarian 
Group, R/GA, & Big Spaceship. http://p.sf.net/sfu/creativitycat-com 
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users