Solaris help ?

[J4CKO]

Need some help with Solaris 10, got some issues and limited skills, after some suggestions as to where I could buy some skills in for a day ?

[J4CKO]

ging84 said:

What does limited skills mean? You have Unix / Linux admins without the skills to confidently deal with some very complex solaris specific issues, or is it more like you have admins who know a bit of nix but lack the skills to confidently deal with anything beyond basic tasks?

Mainly a windows shop, bit of Linux but the Solaris we inherited and the guys that supported that environment have moved on.

My senior DBA has goo Linux skills but for the first time ever he has expressed concern about performing some tasks so would like some insurance int he form of a Solaris expert on hand.

[J4CKO]

Ok, what happened is, we lost a server, a scratch/dev server that hadnt been used in ages, turns out it had the server end of NIS/YP on it.

Question is, can we stop the client end on our Solaris server, what are the implications, we arent sure what will happen if we stop that service, we cant reinstate the server that was removed, we are concerned it has cached info.

I believe NIS isnt Solaris specific, we have a similar server that doesn't have that process.

[J4CKO]

Awesome, many thanks chaps, will do some digging and report back today.

[J4CKO]

This is the contentsof nsswitch.conf on the client



automount: files nis
aliases: files nis

1. for efficient getservbyname() avoid nis

services: files nis
printers: user files nis

auth_attr: files nis
prof_attr: files nis
project: files nis




ypwhich
Domain <domain>.com not bound on <Server in question>.

Edited by J4CKO on Monday 5th June 14:44

[J4CKO]

ypwhich -m

ypwhich(dumpmaps): can't get maplist: RPC: Timed out

[J4CKO]

The server isnt available, this is just on the client.

Also, ls seems to hang my session if I have the e parameter in

Cheers for your assistance !

[J4CKO]

0 383 1 0 40 20 ? 1060 ? ? 986:20 ypbind
0 S 0 5973 383 0 40 20 ? 1060 ? ? 0:00 ypbind

[J4CKO]

gavsdavs said:

J4CKO said:

The server isnt available, this is just on the client.

Also, ls seems to hang my session if I have the e parameter in

Cheers for your assistance !

Oh - so if the server has gone for ever, just remove the nis entries from nsswitch,conf from the relevant maps.

If the server is still there, and staying, then the client isn't bound, and is trying to

What operating system is this nis client ?

Is the server there, or gone and not coming back ?

Servers were inherited, 2 * Solaris servers running VM Zones.

The NIS Server was removed as it was listed as a scratch/dev server used by someone who had left, was part of an aquisition and we got lumbered with it, mainly windows shop, we have Linux folk but not are experienced with this.

It has a twin with the same set up, the etc passwd, hosts etc look the same but no YPBIND running

[J4CKO]

The other server has the same nsswitch.conf file, same entries.

With the files and then NIS, in there, that is the order of precedence, i.e. it checks int he files first then NIS, so if it cant resolve from a local file, then it tries NIS ?

[J4CKO]

gavsdavs said:

J4CKO said:

The other server has the same nsswitch.conf file, same entries.

With the files and then NIS, in there, that is the order of precedence, i.e. it checks int he files first then NIS, so if it cant resolve from a local file, then it tries NIS ?

Correct.

Say you have 30 entries in /etc/passwd, it finds those first, then goes to nis and looks for the passwd map for any extra, 'network' identities.

If the other server has the nis entries in nsswitch.conf but doesn't have a running ypbind, then it's going to behave better as it gets an immediate "not available" from the lack of a ypbind process. (and doesn't get the hang-up effect you're seeing)

What do theese machines need NIS for ? What do they do ?

Lost in the mists of time and the fact the guys who built the environment are long gone, however, have traced the key one, he works for the sister of a current manager at our place so have forwarded some questions to him, via her.

We arent sure what their purpose was originally.

Edited by J4CKO on Tuesday 6th June 09:15

[J4CKO]

seyre1972 said:

J4CK0/GavsDavs,

Just be mindful that until you come up with a replacement for the NIS server - you're relying upon the cached credentials on the running client.

Is there anyway you can salvage/recover the NIS Masters maps/source ? Or is it for the scrap heap ? Is it the disk that failed, or the h/w itself ? If the disk is fine can you not add it to another Solaris server and recover that way ?

Maybe post up what the long term plan is (either recovering NIS, or migration to another service (AD for example)

Is the NIS client a Prod server ? What impact is it if the running server reboots/loses the cached credentials ?

As maybe ad interim either recreate NIS master, or create/add local users/groups/passwds (you could then leave nsswitch.conf alone as the files nis order is fine.

The two Solaris servers host virtual machines with a variety of stuff on, hence the concern.

The NIS server was, I believe also Solaris but was a virtual machine and was deleted as it was thought it was not needed during a clean up.

Long term plan is to stabilise then migrate later operating systems.

[J4CKO]

Guys, stopped NIS tonight after much debate and research, backups seem to be working ok again and the failing commands like "ps -ef" now complete without hanging the session.

So, hopefully we will finally have some valid OS backups and a working machine.

We have also ordered the three disks that have failed in the NAS it sits on, talk about sailing close to the wind fss.

Cheers for all your help !

Quite fancy being a Linux admin, I like working with NIX operating systems, seems so pure after windows.