Bacula-users

Re: [Bacula-users] Regression testing - let's get more involved

2017-05-11 07:45:05
Subject: Re: [Bacula-users] Regression testing - let's get more involved
From: Dan Langille <dan AT langille DOT org>
To: Andrea Venturoli <ml AT netfence DOT it>
Date: Thu, 11 May 2017 07:43:54 -0400

-- 
Dan Langille - BSDCan / PGCon
dan AT langille DOT org


On May 11, 2017, at 5:36 AM, Andrea Venturoli <ml AT netfence DOT it> wrote:

On 05/10/17 14:04, Kern Sibbald wrote:

After thinking about this some more, it seems to me that it is a waste of your time to run Bacula in a jail.  You are going to have nothing but problems.  Running regression tests with Bacula is designed to confine itself to the regress directory (and perhaps tmp), and the probability of having a security problem of any kind while running regression tests is essentially zero.   Sorry, but I would not even try running Bacula regression tests in a jail.

All my Bacula regression testing is done in jails.  The primary goal was to keep the environments separate.  Each regression test I use uses a different version of MySQL / PostgreSQL.  I realized most people only test one version of the DB.  I'm testing multiple because I can. The easiest way to do that is to give each test its own home.


Unfortunately, recent policies do not allow me to try this in base of a production server and that's what I have now.
If doing this in a jail won't work, I'll have to wait until I have a spare machine available and that will be a one-time shot anyway (while I hoped to leave the jail running with crons).

If that's not too much a waste of your time, I'd be curious to know why a jail won't do.

It will do.  It should just work. Bacula has no idea it's in a jail.  I run my production bacula-dir in a jail.  I have several bacula-sd in a jail, some accessing tape libraries.

TIA.




I don't know why it's trying /bin/csh
Ok, so I changed the shell to sh (perhaps this should be picked up
anyway) and I got some further.
However all tests are failing.
Yes, we do not support csh, only sh (and probably bash).

That's fair: I wouln't expect anyone to support csh.
However that's the default on some systems and, in spite of that, I never had the need to explicitly change the shell of the user: the configure script and/or the shebangs would pick sh anyway.
I suggest a note is added to the documentation that everything must explicitly be run from sh.





I used "git clone http://git.bacula.org/bacula.git bacula": that's
different from what the doc says (there's "bacula", not "bacula.git").
Is this ok?
Can you tell me *exactly* where the doc says it is bacula.  The repo is called bacula.git, and if you are using git itself rather than http, it knows to add the correct extension.

I'm using this guide, as suggested by Dan:
http://www.bacula.org/7.0.x-manuals/en/developers/Bacula_Regression_Testing.html

There it says to type "git clone http://git.bacula.org/bacula bacula".
However, on my system:
$ git clone http://git.bacula.org/bacula bacula
Cloning into 'bacula'...
fatal: repository 'http://git.bacula.org/bacula/' not found

So I had a look into http://git.bacula.org/ and run:
git clone http://git.bacula.org/bacula.git bacula

This worked.
I think I grasp the basics of git, but I'm no expert.
If that's a weirdness on my box, please ignore this.

I use git sometimes, and yeah, it's not always intuitive at first.

We no longer support Sqlite.  However, we still keep scripts for Sqlite3.

Sorry, I meant sqlite3-3.18.0.

I *strongly* recommend using only MySQL or Postgresql.

I'll eventually try MySQL (and Postgres when I'll be able).





It is not obvious what is going wrong, but you can get more information by running:
REGRESS_DEBUG=1 tests/<test-name>

I found the problem and was able to run the tests.
Right now I'm at 3 tests failing, but I will investigate these too ASAP.

Some tests just fail.  Review http://regress.bacula.org/index.php?project=Bacula-7.4&date=2017-05-10

The only failures are on MySQL 5.6, not on 5.5, or PostgreSQL 9.[2-6]

Please persist. Having more regression testing is vital to ensuing the next release works and new stuff
does not break old stuff.  The more diversity in the testing, the better the results.

You (the plural you) do not only FreeBSD to represented here.


-- 
Dan Langille - BSDCan / PGCon
dan AT langille DOT org

------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Bacula-users mailing list
Bacula-users AT lists.sourceforge DOT net
https://lists.sourceforge.net/lists/listinfo/bacula-users

ADSM.ORG Privacy and Data Security by https://kimlaw.us