Few observations with ltsp-server 5.9 | Where I can find old ltsp-server 5.1 repository

[Radosław Bursztynowski]

Hi,

Owing to help Joshua (many thanks again) I installed new LTSP K12 on my
server, and I have few observation with it. I will try to do with this
something later, but today my fresh views are as follow.

Let me explain that my old thin clients are very old (the first one:
Pentium III 1000 MHz, 512 MB of memory, GeForce 7300; the second one:
notebook Compaq nc 8230, 2,4 GHz processor, 2 GB of memory, ATI graphic
card), ltsp-server 5.9 and LTSP client epel-6-i386. Both my old clients
cooperated with LTSP 5.1 very good.

Views:

1. Notebook Compacq nc8230 starts booting, but after starting loading
image from TFTP service (I can see information on my screen) I can see
black screen only.

2. Pentium III - boots, I can work with the client, but:
- maximum resolution which I cant reach on my monitor (Samsung
SyncMaster T260HD) is 1680x1050, while with LTSP 5.1 I reached 1900x1200;
- is very slow and watching movies is not acceptable.

I can add, that one my thin client with AMD K6 III 400 MHz, 256 MB of
memory, and GeFforce 5xxx graphic card with ltsp-server 5.1 worked fine,
including movies. I tested LTSP 5.1 with thnin client Pentium 200 MMX, 128
MB of memory, and GeForce 440 and I can say, that this thin client works
with LTSP 5.1 comparable to thin client Pentium III with LTSP 5.9 - not
acceptable. Thin client Pentium 200 MMX was fine with LTSP 4.2.

I will check later LTSP 5.9 with another thin client with Pentium 4 and 1
GB of memory.

But I have old thin clients, which I would like to use now too.

So my question is:

Does anywhere LTSP 5.1 repository exist, which I can use to use my old
think client (let's forget about Pentium 200 MMX, but Pentium III as a thin
client - why not?)?


Best regards,
Radek

[Barry Cisna]

Radoslaw,

I believe all of the "old" K12Linux repos mirrors are out of commission.
I still have a couple servers that are running the 5.1 ltsp server
myself.

Something to consider in your situation/problem. Are you running a gigE
lan with a gigE lan nic in your server? This is pretty much a must at
this stage of ltsp. We used to run only 100mb lan/switches back in the
day of 2.4 kernel ltsp and worked fine.For whatever behind the scenes
stuff thrown into the kernel these days,a gigE backbone is a must.

For kicks i run one client in a test bed on an old 10 base-t HUB. You
wouldn't believe how long it took the one client to boot up!

If you do not have a gigE switched lan you should at least get this
up,and quite possibly your pentium machines will have enough "feed
speed" to work reliably.

BTW. The first versions of k12ltsp we used several 'dumpster' 486
machines and they did work quite well for that day and age :)

Take Care,
Barry

[Joshua Trimm]

In terms of bandwith issues I have included a x2go client as one of the
usable screen's in the ltsp-5.4.4+ rpm's. I apologize for the lack of
documentation on it, I haven't had time to write up proper instructions.
But if you are brave and know bash, have a look at
/opt/ltsp/i386/usr/share/ltsp/x2gothinclient_start (assuming an i386
chroot). X2go uses the same technology that NX uses to do X11
compression. It basically is as simple as using SCREEN_03=x2goclient in
your lts.conf. It's still being developed so customization is purely in
your hands. VirtualGL support is also on my list to deliver 3D desktop
support to thin clients without the need for any 3d graphics cards on
the thin clients themselves.

-Joshua

Post by Barry Cisna
Radoslaw,
I believe all of the "old" K12Linux repos mirrors are out of commission.
I still have a couple servers that are running the 5.1 ltsp server
myself.
Something to consider in your situation/problem. Are you running a gigE
lan with a gigE lan nic in your server? This is pretty much a must at
this stage of ltsp. We used to run only 100mb lan/switches back in the
day of 2.4 kernel ltsp and worked fine.For whatever behind the scenes
stuff thrown into the kernel these days,a gigE backbone is a must.
For kicks i run one client in a test bed on an old 10 base-t HUB. You
wouldn't believe how long it took the one client to boot up!
If you do not have a gigE switched lan you should at least get this
up,and quite possibly your pentium machines will have enough "feed
speed" to work reliably.
BTW. The first versions of k12ltsp we used several 'dumpster' 486
machines and they did work quite well for that day and age :)
Take Care,
Barry
_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>

[Burke Almquist]

Post by Barry Cisna
If you do not have a gigE switched lan you should at least get this
up,and quite possibly your pentium machines will have enough "feed
speed" to work reliably.

To be clear, the clients can be fast (100mbit) ethernet, but the server NIC and the port on the switch that the server is hooked to should be gigE.

As long as we are on the subject ...
Radoslaw, (or anyone else that is interested for that matter) I do have a spare 26 port switch sitting around. It's an AMER SR24G2 (2 gigE ports and 24 10/100 ports). Perfect for hooking a server or two up to a lab full of clients. It's free to anyone that wants it. All I ask is that you pay for shipping.

[Radek Bursztynowski]

Baryy,

Thanks a lot for you answer.
All my computers in the network have 1 Gb/s NICs. I tested (LTSP 5.1) old machines (Pentium III, AMD K 6 III) equipped with 100 Mb/s too. My feeling with 100 Mb/s NICs (still LTSP 5.1) are quite fine. If one or two machines (thin clients with 100 Mb/s NIcs) runs and plays movies, movies are acceptable. If several thin client with NIC 100 Mb/s runs and don't play movies - everything is OK. More thins clients with 100 Mb/s I didn't test. I forgot to say that all my network is equipped with 1 Gb/s switches.

I didn't have any problems with my LTSP clients and network (still LTSP 5.1), including old thin client (AMD K6 III 400 MHz and 256 MB of memory, Pentium III 1000 MHz and 512 MB of memory and 1 Gb/s NICs). My view about theses old thin clients is that movies (1 Gb/s NIC) in Linux environment play movies fine. Worse is with Windows 7 environment (rdesktop or xfreerd). Then movies play little slowly, but the same situation is with more modern thin client. The same situation with Windows 7 is when I use rdesktop or xfreerdp with no LTSP environment - so it is no LTSP problem.

Trucking this way I replaced LTSP 5.1 with latest LTSP K12 and my view I mentioned in previous may post.

Let me add that my server is quite loaded by different services. This machine is 2core Pentium and one NIC 1Gb/s (bridge) and supports 3 virtual machines (KVM) not high loaded. I know that I have few memory (8 GB of memory), but with LTSP 5.1 everything was OK. I monitor usage of two cores and their average utilization - is between 10 - 20% (standard), with very rare picks. This server is my test environment. Next week I will return to my standard environment (modern processors, more memory, two NICs 1 Gb/s in every server) but today I fight with test environment.

Today night I reinstalled my test environment equipped with two NICs. So I am going to return with new post and with new view.

Thanks again for any answer - I will be satisfied with every.

Best regards,
Radek

----

Radoslaw,

I believe all of the "old" K12Linux repos mirrors are out of commission.
I still have a couple servers that are running the 5.1 ltsp server
myself.

Something to consider in your situation/problem. Are you running a gigE
lan with a gigE lan nic in your server? This is pretty much a must at
this stage of ltsp. We used to run only 100mb lan/switches back in the
day of 2.4 kernel ltsp and worked fine.For whatever behind the scenes
stuff thrown into the kernel these days,a gigE backbone is a must.

For kicks i run one client in a test bed on an old 10 base-t HUB. You
wouldn't believe how long it took the one client to boot up!

If you do not have a gigE switched lan you should at least get this
up,and quite possibly your pentium machines will have enough "feed
speed" to work reliably.

BTW. The first versions of k12ltsp we used several 'dumpster' 486
machines and they did work quite well for that day and age :)

Take Care,
Barry


_______________________________________________
K12OSN mailing list
***@redhat.com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>;

[Radek Bursztynowski]

Burke,

I know network limitation. My problem is, that on the same environment with LTSP 5.1 thin clients worked fine. With new LTSP - no. I asked myself - what happened (there is no any changes in switches, cables, server (and NIC), and thin clients (and NICs); I updated CentOS 6.3 to 6.4 and LTSP 5.1 with to newest)?

Best regards,
Radek

Post by Barry Cisna
If you do not have a gigE switched lan you should at least get this
up,and quite possibly your pentium machines will have enough "feed
speed" to work reliably.

To be clear, the clients can be fast (100mbit) ethernet, but the server NIC and the port on the switch that the server is hooked to should be gigE.

As long as we are on the subject ...
Radoslaw, (or anyone else that is interested for that matter) I do have a spare 26 port switch sitting around. It's an AMER SR24G2 (2 gigE ports and 24 10/100 ports). Perfect for hooking a server or two up to a lab full of clients. It's free to anyone that wants it. All I ask is that you pay for shipping.


_______________________________________________
K12OSN mailing list
***@redhat.com
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>;

[Burke Almquist]

It looks to me as if there have been a number of changes in the setup for LTSP on EL/Fedora under the new version. I think the network setup and the build process have changed significantly. Thus we are having significant issues. Some of the fedorahosted wiki information, particularly the setup information that is specific to how Fedora/EL chose to integrate LTSP, may not apply now. I think that is part of the reason for our recent troubles. We really need to get the wiki updated with good information.

Post by Radek Bursztynowski
Burke,
I know network limitation. My problem is, that on the same environment with LTSP 5.1 thin clients worked fine. With new LTSP - no. I asked myself - what happened (there is no any changes in switches, cables, server (and NIC), and thin clients (and NICs); I updated CentOS 6.3 to 6.4 and LTSP 5.1 with to newest)?
Best regards,
Radek

[Radek Bursztynowski]

I begin to think that my trouble can be joined with 2 NICs. It is my
first experience with 2 NICs configuration. The true is that my server
had two NICs, but one of them I didn't configure - was switched off.
Perhaps LTSP 5.1 recognized two of NICs, but I used only one, and LTSP
5.1 didn't force me to interesting of one downed NIC.

Now we can read in LTSP doc that the second NIC receives 192.168.67.1 IP
address by default. So LTSP 5.9 something do with second NIC (perhaps
something different like LTSP 5.1). Finally I could use LTSP 5.9 with my
main NIC, but network worked slowly. This is my surmise only.

I confirm myself with this think because after my new server
installation (server with two configured NICs), I was forced to
reconfigure Samba service, because Samba saw eht1 NIC as default
interface instead of br0, like in the past. When I described in smb.conf
br0 as a Samba NIC - Samba works fine.

But Firefox works to slow. We can inform Samba in smb.conf which NIC
should be use, but Firefox use system configuration. I think that I
should tell system which NIC is the first. But I don't know how.

I repeat that it is my surmise only. Perhaps tomorrow I will be more
clever.

Best regards,
Radek


---

Post by Burke Almquist
It looks to me as if there have been a number of changes in the setup for LTSP on EL/Fedora under the new version. I think the network setup and the build process have changed significantly. Thus we are having significant issues. Some of the fedorahosted wiki information, particularly the setup information that is specific to how Fedora/EL chose to integrate LTSP, may not apply now. I think that is part of the reason for our recent troubles. We really need to get the wiki updated with good information.

Post by Radek Bursztynowski
Burke,
I know network limitation. My problem is, that on the same environment with LTSP 5.1 thin clients worked fine. With new LTSP - no. I asked myself - what happened (there is no any changes in switches, cables, server (and NIC), and thin clients (and NICs); I updated CentOS 6.3 to 6.4 and LTSP 5.1 with to newest)?
Best regards,
Radek

_______________________________________________
K12OSN mailing list
https://www.redhat.com/mailman/listinfo/k12osn
For more info see <http://www.k12os.org>