need help on containers/zones

Hello,

I have been using sparc workstations :SUNW,Ultra-5_10

Total 4 such machines. Users use it for mpich programming and all run solaris 8

but I always have hard time maintaining these machines. Authentication for these machines work from solaris 10 using NIS and there are nfs mounts on each work station.

I am wondering if it would be possible to move these 4 workstations on solaris 10 on V240 by using solaris containers.

V240 is having 8GB of memory. I am wondering if this is possible?

Please guide me , thanks

:b:

It looks doable by using Solaris 8 containers that allows "cloning" the existing servers.

Solaris Containers

Correct, sun has a nice documentation as always. Thank you.

I will be really happy, if someone who has tried this before shares their experience..

:b:

What type of maintenance issues do you have?

Moving them into a container might not resolve your issues, instead making your environment more complex...

I do not want to see " can't open boot device" "disk error" kind of messages. I do not want these work stations shock me suddenly going offline due to any hardware failure. it has been happening regularly with these workstations such type of things..in case of such issues every time I have to take out HDD and try it in different machines, if that does not help then mirror another disk put it in these workstations...my space in the room going waste..these machines need keyboard connected to boot them..i do not have spare ones if something goes wrong with existing ones...more important is system is going down anytime causing loss of work to logged in users...

always these systems go out of swap space, there is limited disk space, /var always around 100%, can't patch them, some days back , it was not hard disk issue but I had to move hard disk to other machine , looks like bad memory on original machine...so such issues...

i don't think i should wait for this hardware to die. Also buying 4 separate sparc machines would be costly. what better way is there than using containers ?

Containers will help to address this type of issue.

"maintenance" can be anything - user issues, environment issues, etc. These issues are not as easily addressed with Containers.

Download the "Solaris 8 Containers 1.0 for Solaris 10 OS, SPARC". Also, grab " Example Solaris 8 System Image" so that you can experiment.

Give it a try!

can these development boxes be solaris 10 or do they have to solaris 8?

Yes, they can be solaris 10, but how and where can I make use of solaris 10 ?:confused:

Thank you all for your replies !

I forgot something to mention but being this an internal network, users use rlogin instead of ssh, I guess that should also be configurable in containers..

Thanks

You can obviously use Solaris 10 in Solaris 10 zones and it is certainly the recommended way. Solaris 8 zones are there to smooth migration but you should eventually run native Solaris 10 zones.

Nothing to configure. The Solaris 8 zones being clones of your previous independant servers, they will behave the same and implement the same services.

i highly suggest you take a look at the documentation for zones under docs.sun.com.

the setup of zones is very easy. however, you do need to plan. resources will be shared (mem, cpu, disk) so this is something to consider. also, you will essentially be running 5 servers (with your current setup) on one physical machine. hardware issues could bring your dev team to a halt. you may consider an upgrade from the v240 you have. if you have a spare box, do some testing. work out your kinks and find what fits for you before you roll it out to production. i've found happy clients (programmers) make for a better day for an admin! :smiley:

i have done this many times and it has worked beautifully.

telnet,rlogin,rsh shouldn't matter as long as you are within your intranet. we are a major financial institution and use those as well... but we are also behind several layers of firewalls :smiley:

Thank you very much for replying with valuable information.

Good day

Thanks jlliagre for this information.:b:

I believe that the v240 should be more than sufficient to replace 4 Ultra 5/10s.

You've indicated that your developers are using mpich - you need to be careful that you aren't making their job more difficult.

Correct, if I could mount the nfs and nis/rlogin working in zones/containers, all I expect is good working environment for users.

Currently the Work stations have 256M or 128M each and users never complained about amount of memory but disk space getting full made ssh /rlogin connections not working/ programs stopped working...

Thanks for your input, even I feel I could give it a try on v240, I feel like it is currently under utilized.. here is current status from top

under-utilized for sure.

avronius - i'm looking at it from a hardware prospective. there are some SPoF issues to deal with. cpu, disks etc... an older model of v240 could pose further issues. i agree that the machine can handle zones just fine. again, just looking at it from hardware.