Ludwig Nussel wrote:

Hi,

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

We have about 20 servers locally, 10 externally and some virtuals running openSUSE - one thing I think needs some improvement is the openSUSE SNMP setup. See e.g. https://bugzilla.novell.com/show_bug.cgi?id=748418 I think the newest HP Support Packs now take openSUSE into account, and presumably they have also moved up the libnetsnmp level, but otherwise it's a bit of a hassle. For the desktop (we have 10-12), it's not so much an openSUSE issue, it's a general issue of adapting to our internal policies. See for instance openfate#312876 and #312871. For my personal use, I could do with improved parental access controls, but I admit I haven't researched it a lot. -- Per Jessen, Zürich (17.7°C) -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

On Wednesday 11 Jul 2012 09:28:15 Ludwig Nussel wrote:

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

Talk to the City of Largo guy, Dave Richards (blog on planet gnome). He manages a lot of thin clients served by openSUSE. WIll -- Will Stephenson, openSUSE Board, Booster, KDE Developer SUSE LINUX GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 21284 (AG Nürnberg) Maxfeldstraße 5 90409 Nürnberg Germany -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

Le 11/07/2012 09:28, Ludwig Nussel a écrit :

Hi,

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

cu Ludwig

Hi ! I've a friend who manages a server infrastructure with almost only openSUSE servers in it. There are 29 servers running. Most of them are replicated in high-available clusters. He uses heartbeat and drbd for clustering. There are all types of servers here : DNS, mail, web, database... He has used the evergreen project (based on 11.2) during a long time, but has swtiched recently to 11.4, as it will be the next evergreen. The most interesting tools for him are YaST and zypper (of course). The release cycle seems to be a bit short for him, though... but that's why he's worked with evergreen, I guess ;-) he even contributes to it :-) Greetings, Agemen. -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

Le mercredi 11 juillet 2012, à 14:25 +0200, Ludwig Nussel a écrit :

Will Stephenson wrote:

...

On Wednesday 11 Jul 2012 09:28:15 Ludwig Nussel wrote:

...

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

Talk to the City of Largo guy, Dave Richards (blog on planet gnome). He manages a lot of thin clients served by openSUSE.

Very good hint, thanks! His blog is here: http://davelargo.blogspot.de/

He found interesting issues, like the dbus daemon running out of file descriptors at ~100 open user sessions.

This is https://bugzilla.novell.com/show_bug.cgi?id=739743 which is still lacking agreement on how to move forward. Vincent -- Les gens heureux ne sont pas pressés. -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

medwinz wrote:

On Wednesday, July 11, 2012 12:08:09 PM Ludwig Nussel wrote:

...

Wow, that's huge! Wouldn't it take you years to take the dvd and drive to each and every school in order to start an upgrade there? Do you monitor all those installations after the initial installation somehow?

Well, we let the teachers manage their lab however we train them how to doing the upgrade.

Ah, ok. Some more questions came into my mind meanwhile (would be also interesting to hear how other people in the audience solve those). Since you use openSUSE Edu I suppose the teachers use iTALC to control the individual workstations, right? Ie monitoring what's on the screen, shutting the computers off if needed etc. Do the children have individual user accounts with their private home or do you use some shared user per workstation? How do you handle printer accounting? I heard that this isn't that straight forward with cups.

We have a small script just to check whether the school server synchronize the e-learning data from our server. Because this setup depends on our wireless/radio network we use the rrdtool and monitor the status of school routers, radio etc.

Anyway, I also use openSUSE for Asterisk in some places in Indonesia. It's a bit old but it runs well :) see this [1] and [2]

And we have an old system with SuSE Professional 9.0 running a software (based on python) to turn-on the sirene on the beach when an authority announce the tsunami warning :) [3]

Hehe, that's cool :-) cu Ludwig -- (o_ Ludwig Nussel //\ V_/_ http://www.suse.de/ SUSE LINUX Products GmbH, GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer, HRB 16746 (AG Nürnberg) -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

* medwinz [2012-07-11 11:23]:

Currently we have 350 elementary and junior high schools that use 21 PC loaded with openSUSE Edu Li-f-e 11.3 and 11.4 in their lab. In total there are around 2300 11.3 and the rest are 11.4. We maintain a local update repository in SLES 11 SP1. 11.3 is outdated now and we think to upgrade it to 12.1 but we still don't find the easiest way to do this actually. We think to upgrade all the PC locally using DVD. The problem is the schools are dispersed and the connection to our repository sometimes broken, we use our own private wireless network. This link http://ict-eqep.com/locations/map_350.html maybe can help you figure out how big it is.

We use the PC for e-learning. We setup Moodle in 1 PC in every schools, synchronize it with the moodle mysql database in our datacenter and pull the e-learning data from data center. Other 20 PC then access the local e-learning material using browser.

Out of curiosity, is there some further technical and/or administrative documentation of this project available apart from the links you already posted? I haven't been able to find any jogjabelajar.org or pemda-diy.go.id. -- Guido Berhoerster -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

Hi Hans-Peter, in terms of management, what are the improvements you would like to see in openSUSE that would have a bigger impact in your everyday work as administrator of this deployment? On Monday 23 July 2012 19:48:35 Hans-Peter Jansen wrote:

Dear Ludwig,

Am Wednesday 11 July 2012 09:28:15 schrieb Ludwig Nussel:

...

Hi,

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

I'm responsible for an installation with 30 clients and two servers in a transport company, everything is (still) openSUSE driven (apart from a bunch of VMware XP vm's running ontop).

_All_ clients are diskless, a kiwi setup is tweaked to cope/use the aufs2 patched kernel. It's a real pity, that the kernel team opted for outphasing aufs a long time ago. It's still the stablest solution, when it comes to run nfs3 based diskless full featured desktops (including XP under VMware).

The client distibution is installed with a python script, that runs zypper, applies some patches, and prepares the clients layer. The installation target is a nfs exported directory on the server, each client runs on his own rw filesystem layer ontop. That way, I can easily manage/modify every client on his own without affecting others, while still being able to apply updates in a single place once and for all.

Since we use multihead setups (up to _3_ * 24' heads), we selected nvidia as our favorized gpu, driven by the closed source drivers (a decision made back in 2005).

A single, but pretty important win32 module is driven by wine to access a transport trading exchange (Timocom).

We still use 11.1 and KDE3. Unfortunately, the KDE team has lost track, when it comes to business users needs (imap is not integrated in desktop search engines, kaddressbook is neither company aware (one company to many individuals, nor branch aware), and lacks a sql database backend. kmail is full of minor issues in such a setting.

We had a couple of very nagging LibreOffice issues, and still have some.

Let me note, that this setup is the third one in a row, both predecessors were SuSE based, too, and the immediate one was aufs based already. The first one dates back to 1999.

It biggest advantage of this setup is flexibility, while managing it is pretty lightweight. Adding or exchanging systems is done in a few seconds. Automatic X setup is flaky, hence multihead setup has to be done manually. I fear, that the sax2-less versions will further complicate this issue.

Changing things has to be done very conservatively, since such an environment doesn't tolerate a lot of experimenting, and time is constrained. For that reason, the short release cycles of openSUSE are a real pain. OTOH, the advantage of the professional desktop products isn't that big, given the many deviations from the original product over time.

Back in 2.4 times, we had setups, that were rock solid - desktops running diskless for month without reboot, including VMware vm's.

Generally, I notice decreasing long term stability in a lot of areas, that really disturb me and that prepare the fields for Redmonds takeover.

Cheers, Pete -- Agustin Benito Bethencourt openSUSE Team Lead at SUSE abebe@suse.com -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

Dear Agustin, well, call it the usual suspects ;-) e.g. stability and long term commitment. Check out Ubuntu LTS projects to get an idea. Please don't get me wrong - I'm still commited to openSUSE in every respect. On a large part due to me being here since 1997 at least (as one of the SuSE beta tester), the KDE(3) support, and finally, I don't like the debian way of building packages. But I do note, that the openSUSE project is cleft in so many small areas, that in the end, it is nearly unpossible to concentrate their attainments in stable products again. Linux would be nowhere today without some kind of benevolent dictatorship, and that seems to be missing here (well, sure, controlling a software project is easier than controlling a distribution of some such.., especially when driven by community effort at large). Shuttleworth seems to get this right. While at it: I really appreciate the deferred release of 12.2 - and would see this as a call to the participating parties to relax the release policy in a way, that is able to finally raise stability of the released product. I still have one asterisk system running on top of the unbeatable solidness of SuSE 9.3 with a current uptime of 787 days and 2 hours. Well, these times are gone. Today, the development pace of many important projects is simply too big, where coding is done much faster, than testing of many aspects is even possible. In the middle and long run, some kind of artifical deceleration needs to be established, and more conservative rules in order to allow publishing stable releases. Those, that want/need to live on the bleeding edge will choose rolling release products/distributions anyway. Pete Am Thursday 26 July 2012 11:07:23 schrieb Agustin Benito Bethencourt:

Hi Hans-Peter,

in terms of management, what are the improvements you would like to see in openSUSE that would have a bigger impact in your everyday work as administrator of this deployment?

On Monday 23 July 2012 19:48:35 Hans-Peter Jansen wrote:

...

Dear Ludwig,

Am Wednesday 11 July 2012 09:28:15 schrieb Ludwig Nussel:

...

Hi,

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

I'm responsible for an installation with 30 clients and two servers in a transport company, everything is (still) openSUSE driven (apart from a bunch of VMware XP vm's running ontop).

_All_ clients are diskless, a kiwi setup is tweaked to cope/use the aufs2 patched kernel. It's a real pity, that the kernel team opted for outphasing aufs a long time ago. It's still the stablest solution, when it comes to run nfs3 based diskless full featured desktops (including XP under VMware).

The client distibution is installed with a python script, that runs zypper, applies some patches, and prepares the clients layer. The installation target is a nfs exported directory on the server, each client runs on his own rw filesystem layer ontop. That way, I can easily manage/modify every client on his own without affecting others, while still being able to apply updates in a single place once and for all.

Since we use multihead setups (up to _3_ * 24' heads), we selected nvidia as our favorized gpu, driven by the closed source drivers (a decision made back in 2005).

A single, but pretty important win32 module is driven by wine to access a transport trading exchange (Timocom).

We still use 11.1 and KDE3. Unfortunately, the KDE team has lost track, when it comes to business users needs (imap is not integrated in desktop search engines, kaddressbook is neither company aware (one company to many individuals, nor branch aware), and lacks a sql database backend. kmail is full of minor issues in such a setting.

We had a couple of very nagging LibreOffice issues, and still have some.

Let me note, that this setup is the third one in a row, both predecessors were SuSE based, too, and the immediate one was aufs based already. The first one dates back to 1999.

It biggest advantage of this setup is flexibility, while managing it is pretty lightweight. Adding or exchanging systems is done in a few seconds. Automatic X setup is flaky, hence multihead setup has to be done manually. I fear, that the sax2-less versions will further complicate this issue.

Changing things has to be done very conservatively, since such an environment doesn't tolerate a lot of experimenting, and time is constrained. For that reason, the short release cycles of openSUSE are a real pain. OTOH, the advantage of the professional desktop products isn't that big, given the many deviations from the original product over time.

Back in 2.4 times, we had setups, that were rock solid - desktops running diskless for month without reboot, including VMware vm's.

Generally, I notice decreasing long term stability in a lot of areas, that really disturb me and that prepare the fields for Redmonds takeover.

Cheers, Pete

-- Agustin Benito Bethencourt openSUSE Team Lead at SUSE abebe@suse.com

-- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

On Fri, 3 Aug 2012, Kim Leyendecker wrote:

BTW, imho, software should be released when it's done, not when eight months are over and the schedule complains for a new release.

By virtue of human nature, there may be a relationship between setting a goal and being ready at that time (plus delta). "It's ready when it's ready" does not seem to work too well for free and open source projects, especially larger and more complex ones. Gerald -- Dr. Gerald Pfeifer , Director Product Management, SUSE -- To unsubscribe, e-mail: opensuse-project+unsubscribe@opensuse.org To contact the owner, email: opensuse-project+owner@opensuse.org

On Wednesday 11 July 2012 14:25:46 Ludwig Nussel wrote:

Will Stephenson wrote:

...

On Wednesday 11 Jul 2012 09:28:15 Ludwig Nussel wrote:

...

I'm currently investigating whether and how openSUSE is used in larger deployments. Does anyone know someone who manages like let's say 30 or more openSUSE machines at once? I'm interested in e.g. what tools are used in such a scenario and what rough edges in the distro we maybe have that could be improved.

Talk to the City of Largo guy, Dave Richards (blog on planet gnome). He manages a lot of thin clients served by openSUSE.

Very good hint, thanks! His blog is here: http://davelargo.blogspot.de/

He found interesting issues, like the dbus daemon running out of file descriptors at ~100 open user sessions.

cu Ludwig

Please also try to document everything you find on: http://en.opensuse.org/openSUSE:In_use ;-)