What I Learned Today

netdom query fsmo

gives you a list of all the fsmo role holders in your domain.

i don't understand quite what is going on with our cvs server's mail config - seems that both sendmail and qmail are installed. Both Qmail and SendMail are logging info into the syslog, a majority of it being sendmail info.


**************
Testing an SMTP server via the command line

telnet smtp.smtpdomain.com 25

ehlo somedomain.com

mail to:someuser@smtpdomain.com
rcpt to:someuser@somedomain.com
data
subject some subject

CTRL D CTRL D

quit

*************

1. I had to setup our Knowledge base to work externally as well as internally and I completely forgot that I needed to setup a listener on port 80 on the firewall for the IP address that is associated with the DNS name. I had setup the destination set and the web publishing rule but not the listener. Once I turned that on, all is well.

1. More AS/400 fun - I finally replaced the bad disk tonight using the handy dandy - AS/400e Server 170 and 250 Problem, Analysis, Repair and Parts guide. its a 600 page book. i only needed about 5 pages though and it was fairly straight forward. So for $50 for the replacement hard drive, I was able to get the system back up to full strength. But now it is complaining about auxillary storage shortage... so i'm not sure what all is going on.

i'm a bad bad blogger... how am i going to remember what I learned if I don't write it down.

1. Still dealing with the AS400 and the bad disk. I decided I should do a full backup of the entire system that way if I end up destroying data when I remove the bad disk (despite the OSs reassurance that I won't) I can atleast attempt to restore from a full backup. Turns out a full backup is a pretty big deal and requires shutting down all the subssystems. I ran the backup on Thursday before I left at 4:30. I was out of the office next day and naively assumed the system would bring it self back online after it completed the backup. I was wrong. Since I was en route to KY on Friday, I was stuck on the phone trying to walk poor Mo through bringing the system back up. It never did come back online so it was waiting for me on monday. It wasn't a big deal to bring it back it up, it seems it was in "attended" IPL mode. Once the system was online, all was well. Every Monday night the machine automagically IPL's itself. Which has never been a problem before, except that I have to mount the NFS shares on Tuesday morning. Well yesterday, i went to mount the shares, but the system wasn't up. It seems the system was waiting for assistance and also was in "attended" IPL mode. I'm not sure if this has to do with the bad disk, or something that happened with the back or something that Mo inadvertantly did.. but its rather annoying. The plan is to now replace the disk on thursday evening (we are off on Fri).

2. Currently I am using backup exec for the backup of all my dev systems. This works ok, but using a single DLT drive is a pain and it is not part of the weekly offsite tape rotation. Everyone keeps saying how amazing rsync is.. so I figured I would set something up using this little tool. I need to backup everything but the NFS mounts, so I'm doing something like this...

rsync --single-file-system -aRv / /usr /var /tmp helios:/nas/qaroot/software/images/nobackup/'hostname'

yes i'm using .. cringe .. rsh .. but I have my reasons.. mainly. i'm lazy.

3. I have still not figured out my mac issues.. nor why my new NIS slave doesn't seem to be working quite right..

4. There is a big buzz about Apple's bootcamp -- its pretty slick allowing you to boot windows on an intel mac. There is also virtualization software out there that will let you run windows simultaneously with OSX. What would be interesting to me would be software that would allow you to buy a mac box, load with windows, then run OSX in a window...

5. Beth is so good at documenting my travels... she is pretty creative too , she used the camcorder to record the radio broadcast and that made it easy for me to transfer it to the computer. Here is a pic from inside the big empty, black ride...

1. I'm learning more then I ever wanted to know about the as/400... We need a new one. So I went down the normal route of using IBM's developer program to get us a new box with a discount - but that ended up costing about $75K ... and considering we have about 2 customers on the platform, it doesnt make sense to spend that much on the hardware. So I went down the used route and found some of the old school (ie not a pseries) hardware for incrediably cheap. Currently we have a 170-2292 ( 220 cpw / 30 cpw interactive) and I'm looking at buying a 270-2434/1516(2350 cpw/ 70 cpw interactive). What I haven't figured out yet is if the extra horse power is going to help us at all. We only have two developers at most working on the system and they use it for interactive compiles our of product and they also use it for interactive compiles of our sample applications that use our product. There is no other work being done on the machine. I looked in the QSYSOPR message queue and didnt' find that we were running out of interactive cpw so I am unsure if we are even processor bound on the machine. If we aren't then I should probably spend the money on additional RAM and disk space as opposed to interactive cpw.

2. I also learned that we have a disk problem on our existing 400.. I was looking in the QSYSOPR message queue for any indication of interactive cpw issues and found a boat load of disk error messages. It seems that I have disk in the array that has died.. oh joy.

3. Last monday I found out that our building is killing the power on sunday morning (tomorrow) for eight hours ! Eight Hours ! Who has a battery powered UPS that can last for eight hours??
Since our website is actually located in my server room I was able to transfer it over to our ISP's hosting service and then get the DNS records updated for the weekend so that people will end up there instead trying to access my powerless "data center" (i use the term loosely). Our ISP also has an mx record for our domain so they will hang onto the incoming email while the exchange box is powerless. So all that is left for me to do is gracefully shut down the 25 development servers and 17 infrastructure servers, NAS, phone system, security system, and then go in Sunday afternoon and turn them all back on. Lucky Me !

1. I've move from interop hell to solaris patch help -- we have a customer that wants us to port our app to Solaris 64-bit w/SunStudio 10. Of course SunStudio 10 costs 3K even though SunStudio 11 is free. Anyway - I get the eval version of the product and install in on the our Sol 8 box. The developer runs into problems immediately and asks if all the compiler patches have been applied. So I checked the download page and there was no mention of additional patches available... curious. So I search the site and find the patch page for Studio 11.. but not ten. Turns out the magic search phrase was Sun Studio 10 patch page That lead me to a page that contained links to patches for all the Sun Studio products. This is where patching for Solaris / Sun studio starts to stink pretty badly -- compared to AIX and HP-UX.. and even red hat for that matter. I found a grocery list of patches that can be applied, each for my individual downloading and installing pleasure. There is no ability to bundle the patches, you simply must download the .zip file and copy to the machine and install it. So I start downloading the 15 or so patches that were recommended. Here is how I was able to "batch" install the patches

1. download patches to NFS mount
2. cp *.zip /tmp/patch ; cd /tmp/patch
3. unzip \*.zip ( the \ is required when using a wildcard for unzip ... doesn't work otherwise.)
4. ls > patches (unziping the patches creates a directory with the patch name for each patch, kicking those directory names into a file creates the list of patches we want to install for the next step)
5. patchadd -M /tmp/patch patches

its not pretty but it works --


2. I was thinking about my Mac interop problems and maybe I don't need OD at all. I was thinking I could just leave the user profiles local to the machine. So that way i wouldn't need "managed options" or whatever those mac fellas call it. As it stands now... the Xserv is still hosed up though - when i attempt to connect to an smb share on the box with a windows machine, I get "The account is not authorized to log in from this station" and when i try to connect via smb to the share with a mac client I get "invalid user name and password" . However, I can logon to the Xserv with my windows user name and password, so i know it is bound to the directory correctly its seems to be more of a samba issue.