[Bug 531252] Review Request: lfc - LCG File Catalog (LFC)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug.


https://bugzilla.redhat.com/show_bug.cgi?id=531252

Mattias Ellert <mattias.ellert@xxxxxxxxxxxx> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED

--- Comment #17 from Mattias Ellert <mattias.ellert@xxxxxxxxxxxx> 2010-04-03 12:42:25 EDT ---
(In reply to comment #12)
> Given
> 
> diff /etc/lfc-postgres/lfcdaemon.logrotate /etc/lfc-mysql/lfcdaemon.logrotate
> 
> shows they are the same having an alternative between them seems
> a bit pointless but then of course I realise there is the problem
> of where you put it. Perhaps a -common package or something.

There are in total 7 logrotate files. If each of these should be put in a
separate subpackage this would mean 7 new subpackages each containing a single
file. I don't find this to be a good idea.

(In reply to comment #13)
> In the README change:
> 
>  mysql> grant all on cns_db.* to <user>@localhost identified by <password>;
> 
> to
>  mysql> grant all on cns_db.* to <user>@localhost identified by '<password>';    

Accepted.

(In reply to comment #14)
> Related to comment #12 I think the /etc/NSCONFIG should really be present
> and owned especially given it is in /etc/
> 
> It could go in lfc-common package.    

The /etc/NSCONFIG file contains the username and password of the lfc database
that the sysadmin creates when the service is deployed. This file should not
exist before the database has been created. Its presence is an indication that
the sysadmin has initialized the necessary database tables. It can therefore
not be installed or created at the time of package installation.

(In reply to comment #15)
> 1) Can you submit a bug for 
> # rpmlint dpm
> dpm.x86_64: W: shared-lib-calls-exit /usr/lib64/libdpm.so.1.7.4
> exit@xxxxxxxxxxx
> 1 packages and 0 specfiles checked; 0 errors, 1 warnings.

I sent an e-mail to the upstream developers about this issue and it has been
fixed in the latest version.

> 2) Request only: Given there are two different versions of rf* commands out
>    there any chance these ones could become of managed by alternatives key of
>    rfio so as to exist with.
> 
>      > rpm -ql castor-rfio-client
> /usr/bin/rfcat
> /usr/bin/rfchmod
> /usr/bin/rfcp
> /usr/bin/rfdir
> /usr/bin/rfmkdir
> /usr/bin/rfrename
> /usr/bin/rfrm
> /usr/bin/rfstat
> 
> and man pages. Obviously castor-rfio does not exist in Fedora but it might
> one day. It certainly does here of course.

I sent the following question to the upstream developers:

"The dpm-clients bundles the set of rfio-clients. These are also present in
another software package you provide (castor-rfio). This package is not in
Debian or Fedora yet, so it is not a packaging conflict in the distributions at
this time, but it is a potential one. Do you have any plans w.r.t. this so that
you only provide one set of the rfio client tools. Looking at the sources the
two versions seems to be different versions of the same software a bit out of
sync with each other, not two different implementations that both should be
packaged, but please enlighten me."

And got the following reply:

"There was 3 years ago a project to have a common RFIO between CASTOR and DPM.
But not much has been done and the 2 RFIO codes have even more diverged.
However in the medium term CASTOR wants to drop RFIO. So I'm not sure we should
spend time on this, but the RFIO code cleanup I've done for DPM 1.7.4-2 would
certainly help."

I suggest keeping the current structure of the package until we know more about
whether upstream will merge the two implementations, or drop one of them, or
not.

> 3) I realise this is an upstream thing but both lfc and dpm-nameserver use a 
>    database called cns_db by default. If possible/easy could one be changed.

Upstream's reply:

"cns_db is the default but as you can specify the DB name in the NSCONFIG and
DPMCONFIG files, I do not see really a problem. We have also a plan to be able
to use the DPM Name Server as local LFC (DPM 1.7.6)."

Given this reply, I suggest upstream's defaults are kept. It is tricky to run
both an lfc and a dpm-nameserver on the same host anyway since they both by
default listen to the same port (5010).

> 4) Finally same comments for dpm and lfc about the use of a -common package
>    for each with configuration files to avoid things like.
> 
>    # rpm -qf /etc/DPMCONFIG 
>    file /etc/DPMCONFIG is not owned by any package

The /etc/DPMCONFIG file contains the username and password of the dpm database
that the sysadmin creates when the service is deployed. This file should not
exist before the database has been created. Its presence is an indication that
the sysadmin has initialized the necessary database tables. It can therefore
not be installed or created at the time of package installation.

New version:

Spec URL: http://www.grid.tsl.uu.se/review/lcgdm.spec
SRPM URL: http://www.grid.tsl.uu.se/review/lcgdm-1.7.4.4-1.fc12.src.rpm

-- 
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
_______________________________________________
package-review mailing list
package-review@xxxxxxxxxxxxxxxxxxxxxxx
https://admin.fedoraproject.org/mailman/listinfo/package-review

[Index of Archives]     [Fedora Legacy]     [Fedora Desktop]     [Fedora SELinux]     [Yosemite News]     [KDE Users]     [Fedora Tools]