[K12OSN] OT: still have sluggish system

[dahopkins at comcast.net]

> 
> The high bandwidth usage is an obvious candidate. Any idea on what is
> chewing up the bandwidth?

Not really, usage patterns/programs for the teachers/students haven't changed much year-to-year.  Software packages are the same as last year.  Only change was FC5 upgrade. (mb is Supermicro X5DP8-G2 in case that matters somehow)

> 
> I would also check for authentication/authorization issues (I note in
> the archives that you use LDAP).

Yep, used David Trask's instructions originally, updated with Matt's script later.  Nothing changed though in last year.

> 
>   Double-check that all of the LDAP configs are correct.
> 

Everything is as it should be.  Haven't upgraded to latest samba though because of concerns with the latest/greatest version of samba and the traffic on the samba list about it.

>   Make sure that nscd is running (/sbin/service nscd status)
> 

Have had an ongoing issue with nscd since like forever.  I am using a dual-Opteron server with the prior version of K12LTSP running (x86_64).  Fully patched, but I have not upgraded the system to the FC5 version.  nscd still dies 2-3 times each day.  A simple restart, or at worse clear the database files and restart resolves the issue (on site tech teacher has a script that he runs if he sees the authentication stop). I am planning to move my LDAP server to a new 32-bit system just to get away from this issue.

>   Make sure that DNS is correct (/etc/resolv.conf)
> 

Haven't touched this either, but will recheck.

> 
> I'd also check the disk sub-system...

I did upgrade from Ultra160 IBM Deathstar drives (they were dirt cheap and now I know why) to Ultra320 Seagate Barracuda's.  Was using a zero-channel controller for RAID-1 on the OS drives, but ... that is another thread. :)  Now using software RAID-1 on one system, nothing on the other (skating on thin ice).  Both systems show the same issues.

> 
>   Run "dmesg" and look for SCSI errors
> 
Haven't seen any so far.  Checked earlier, will check again.

>   Run "sar -b" to see the disk I/O usage patterns
>   (if you don't have "sar", install it with "yum install sysstat")
> 
have to install and test this as well.  Only issue is that I am running out of annual leave to spend at the school troubleshooting this.  The issues only show up under load.  Still, will be there tomorrow morning bright and early since it is my day off :)

Sincerely, and thankfully,
Dave ...