CUPS Model NAS oracle Sharepoint

To domain or not to domain

This past couple of weeks has been interesting. My client that I do a lot of work for has decided that its time they moved to a larger supplier. I’ve been with them for four years now, the last two working from my own office. Its tough being a small business supplying services to a very large one. The difference in my costs to that of their outsourced IT is huge, but would not make a dent on their expense sheet.

This past month I have createing some very shophisticated reports for my client. I have been fiddling with the Oracle model clause because these calculations require inter-row calculations. So to cut down on self-joins I gave it a try. Works a charm on my Oracle 11g server with 4gig ram and oddles of temp and swap space. Did work so well on my client’s 10g server with very resrictive limits. It would run out of memory and also take ages to run, locking up the machine. Luckily I was using my client’s UAT server and not production. I had to revert to using self-joins and a full-outer-join to join the two rows I needed to complete the calculation. Test and test and test again I say.

I’ve managed to finish the domain setup on my Linux machine and now am looking at ways to copy the large amount of data on the current NAS, over to the new NAS. The new NAS has a RAID-5 setup with 900Gig of storage and is properly partitioned to store user files, databases, LDAP database, e-mail etc. I used the Linux LVM to manage this system. Is that a bad move? I could not see an easier way to do it. I suppose I was being lazy. The ability to enlarge the individual logical partitions should come in handy down the road however. Let see.

Before I can move all my Windows users on to the domain, I need a way to preserve their profiles on each machine. I am not using the roaming profile available. I have experience from my client site that these don’t work well as you’l find that machines are setup with different software than your profile and so you usually end up using the same machine each day anyway. I’ve found two links to this.

The second way looks to much simpler. I would appreciate any experience people have had with this.

CUPS might be the standard printing for Linux, and maybe some Unix systems. But I struggle with it. Our Konica-Minolta Bizhub 250, a professional MFP, seems to cause me no end of trouble. The main problem is printing letterhead. The usual way to do this is to put the letterhead into the bypass tray and inform the driver that the first printed page will come from the bypass tray. This avoid the need to open the paper trays and manually insert the letterhead there. This works fine from the Vista machines using the Vista printer drivers. But it completely fails from the XP machines. Cups reports the job as ‘printed’, but the printer never receives the job and nothing is printed. Rather annoying and a common problem with CUPS. It simply does not know if the jobs was actually printed or not.

Its an area that needs serious improvement IMHO.

So when I get all the files from the current NAS and split them and place them on the new NAS and reassign the permissions, which will be LDAP groups and users, the old NAS will become a backup server. I’m only thinking of leaving CUPS, SFTP, NTP and maybe DHCP on the current machine. I’m looking for ideas on backup software to run on this machine. I’ve heard of PCBACKUP, but have yet to install it.

This month I have promised my self that I will install knowledgeroot. And as my contract is running out, I will dedicate myself to learning and playing with sharepoint. I think next time I may describe the folly of uninstalling it, installing SQL Server 2008 and re-installing Sharepoint. I’ll need luck and sweat I think.

Then I can move on to Sharepoint consulting, but its a chicken and egg thing. Would a customer hire me – with 15 years general experience but not certified in Sharepoint – or a certified shop. I think I’ll have to offer some loss-leaders to build up my portfolio. Any tips or leads appreciated.

Alfresco CMS DMS Plone Sharepoint

Is it a CMS or is it a DMS

Since my last post where I mentioned that I am looking in the document management field I began to explore other thoughts. Specifically, whether small businesses could use a beefed up CMS (Content Management System) rather than a full blown DMS (Document you guess the rest).

Its my thought that certain systems like Drupal, Plone, dotCMS that have document management modules and collaboration and extension facilities could be used as poor man’s document management system. It wouldn’t have all the bells a whistles like digital signatures, records management with proper archival and other facilities in most of these systems. But, hopefully, such systems could be upgraded to a full DMS if needed. But we’ll cross that bridge when we get there.

There’s always the big bully boy Sharepoint. The WSS seems very bare-bones but right off the bat it makes for very simple document storage. The only hitch being that everything needs to be written in .Net and it requires specialized hosting. I’m going to concentrate on Sharepoint but only so far as application construction is concerned. I reckon that any company well off enough to afford sharepoint and the licencing can afford their own IT department to support it. I’ll be investigating Sharepoint more as time goes on. I’ll be playing with from the point of a small application which I feel is much needed for legal firms. At least it will fill my time while I wait for my next job.

There are the Open Source alternatives – some would call them preferatives (is that a word?). Nuexo, Alfresco, Knowledge Tree – the ECM systems. Built from the ground up a collaborative document management systems. I like these systems, but in terms of price and features. Any of these can be had for free. My company model would be install and support these system on my own with my own support contract with my clients. I would, of course, build in my own bits to hand to my clients only. Oh the wonder of the GPL.

I’m continueing to ponder this. I’m sure after watching some demos I’ll get convinced one way or the other. I do notice that it is a lot simpler to host PHP and MySQL based products than Zope (Plone) or Java products. Should I care if the system will be hosted on the client’s own machines or on a dedicated server? hmmmm….

CUPS DM document management ldap Linux PDC SAMBA Sharepoint

Long train take me home

Hello all,
I’ve finally decided to continue this blog after several months of letting it rot. I’m currently on the bench as they say in the UK, On the beach as they say in North America. A lot has changed in the office and the future work I hope to do, so lets get on with it shall we?

I’ve figured out the CUPS server and all is running very well. My Windows clients have little issue printing to any of the three printers in our office. I only have one small issue with the Konica Minolta MFP where every once in a while I send it a job and the job doesn’t come out. I check with CUPS and it labels the job as “complete” – eventhough it hasn’t printed. I haven’t done too much debugging into this. I’d like some tips as to why that would happen. I believe with older versions of CUPS this was quite common. There are many people on forums indicating that even rebooting didn’t solve their problems, or only solved them for a short period.

Now that I have sorted out CUPS for myself, I’ll slowly be rolling it out to the remainder of the office. One by one I’ll be switching the printers on the PC’s to use the CUPS server rather than use a direct link. There will always be a fall back so I don’t think there’s much risk. I think for the Konica Minolta I’ll leave the direct link on each PC in case the not-printing problem occurs. I’d hate to have user’s scream at me that this new print server is terrible.

I also made significant progress with Linux. I have setup a new NAS server with 900MB of total storage using a RAID 5 consisting of 4 320GB drives. I’ve manged to get it use a volume manager so in theory I can re-size the partitions I’ve created should I need to. The way I have set it up I have left a lot of unused space so re-sizing currently means just expanding the one partition that has filled up.

I then set about putting LDAP on the server and getting it ready for use as a Samba PDC. This was not easy. I searched around and there are few places which give tutorials on this. But its all in bits and pieces. First you need to follow to get OpenLDAP up and running. Then follow, which gets Samba and LDAP talking to each other. You then need to get the authentication to work by following That didn’t work for me, I still don’t know why. I had to additionally edit my /etc/ldap.conf with mappings to map to the correct place in the LDAP hierarchy.

nss_base_passwd ou=People,dc=dom,dc=local?one
nss_base_passwd ou=Machines,dc=dom,dc=local?one

The second line is most important as my PDC would not allow new machines to join without it. This line took me two days to discover and only through a forum entry by someone with a totally different problem than me.

Once I got that going things went swimmingly! I just have some account clean up to do. Now that I have domain accounts some redundant accounts on the machines need to disabled and changed to use the LDAP accounts instead. This is a faily minor procedure on the Linux machines which just involves disabling accounts and performing chown and chgrp with the proper ID’s. On Windows its a bit more complex, though I hear it can be done from the profiles dialog. Apprently you just reassign the profile folder to the local account and then delete the local account without deleting the profile folder. All done.

I’m also rolling this out on a machine-by-machine basis and will hope to have this sorted within the next month.

I’ve been playing recently with the MODEL clause of Oracle SQL. This was introduced I believe in 10g and has been inproved in 11g. One of my clients needs a sophisticated financial calculation performed which used intra-row calculations. The process also needs some “gap-filling” to accomodate somewhat user entry errors. Where there is data missing in one month which was present in the previous and following months. The MODEL clause can do this quite easily, without having to do all sorts of self-joins and left and full outer joins. I was impressed with the elegance and speed of the solution. Now I just need to squeeze it into the tiny operating confinments of the client’s server.

I used to think that disk space was cheap and when the sys admins saw that space was filling up, they would just allocate more from the SAN and when they needed to, buy more disks. Really a 1TB disk is not a lot for a large company is it? Yet my client is constantly plagued with out of space errors on their servers.

I’ve now been pondering the future of my business. Document management seems to be the future these days. With all the requirements for freedom of information and regulatory requirements for financial institutions coming, it seems a growth area. Even in the legal area, there are masses and masses of documents. I can tell you from experience that hunting for template documents and people’s documents can be a nightmare. Even more so if they have been misfiled!

Document management systems can be very complex but there are some Open Source solutions available. Providors like KnowledgeTree and Alfresco are well known in this area. Sharepoint is another obvious one. Though sharepoint is nearly useless in its bare form in my opinion. It needs a lot of work to come up to the level of a Alfresco.

I’m pondering this area and will decide on my future shortly. I’ve also need to consider how to train and market this new offering.

All for now.