jump to navigation

Spamassassin on RHEL 4 (disable per user settings) April 6, 2008

Posted by daakeung in E-Mail, IT, Linux.
Tags: , , ,
add a comment

Recently I have been working with a group, to move their email system from a shared solution to a dedicated server.

Setting up Spamassassin has yield some ups and down.
By default, Spamassassin has per-user settings enabled by default. So any configuration I was making via /etc/mail/spamassassin/local.cf wasn’t being applied properly, as they were over-ridden by user_prefs located in ./home/(user)/.spamassassin/

To disable spamasssassin from using per-user prefs, you need to set a parameter when starting it.

File you need to edit is /etc/sysconfig/spamassassin
# Options to spamd
SPAMDOPTIONS=”-d -c -m5 -H”

The option you need to add is -x

so the output look like this
# Options to spamd
SPAMDOPTIONS=”-x -d -c -m5 -H”
Save and restart spamassassin

service spamassassin restart

and you will be working only with the config located at /etc/mail/spamassassin/local.cf


VMWARE ESX Server slow write when using NFS as a Data-Store. April 4, 2008

Posted by daakeung in Linux, Virtualization, Vmware ESX.
Tags: , , , ,

I’ve been playing around with virtualization for quite sometime not, at the beginning I wasn’t too interested. In fact a friend of mine, Q hooked on this thing like a junkie on crack…heh…. need less to say, I’ve been testing HA, DRS and Vmotion stuff out.

To have all your VMs accessible to your esx servers at any time, you would need a common storage amongst them, the Data Store is this case.

Any of you tested with NFS, would notice because of the sync being done when writing a file to the nfs store, dramatic decrease in throughput for the data, can result in VMs crashing and sometimes in Windows, disk errors show up in the event log.

I currently run Centos 5 (latest update)

/VMWARE-HOSTS x.x.x.x(rw,async,no_subtree_check,nohide,no_root_squash) x.x.x.x(rw,async,no_subtree_check,nohide,no_root_squash)
I was reaching throughput of 9 – 16 mbit with sync, using async throughput has increased.

Now I get more than 80mbit.