Month: March 2008

Script updates
I had to do some update on the weblog script, because the vmswiki and the standard wiki-help paths wer not yet included. I took the opprotunity to some more enhancements: One annoying problem I found on some entries was a warning of a commandline being too long. Researching the issue, I found it was on writing a line to the log, that exceeded 256 characters. By adding “/SYSMBOL” to the WRITE command solved that problem. Still, I cut lines if ther exceed a certain size – just to be sure.
I also added a log line showing the file processed – handy in case of problems 🙂
The adapted script has run against all log files without an unexpected message.

WordPress 2.5
is out for a day now, and just to see how it would behave on OpenVMS, I downloaded the zipfile, and installed it. The basic requiremenst are humble and should fit the PHP engine (4.3) and MySQL (4.1) running on OpenVMS.
Installing required doing a bit of renaming – and that impacts all 2.2.3 blogs on the system: This one, “Trips, Tracks and Travels”, and the old bootcamp-2007 blog; the bootcamp-2008 blog runs WordPress 2.3 so that one wasn’t affected 🙂 by renaming wordpress.dir to wp223.dir. All logicals of the affected blogs needed this minor change included, and the operator entry page, that links straight into the basic, code-containing directories needed to be updated accordingly, as well as the file shat defines these logicals. Just a few minutes and the new definitions were implemented where required.

But once the new WordPress.dir was created and populated – and renamed to wp25 – the protections set properly, the configuration files fitted to fit MY requirements (have the access details out of the way) and an empty WP25 database created, the test blog could be accessed, the database got populated and I could login with the supplied , generated password. The result was a very, very basic page, without any formatting and scheme applied. The reason of this is simple: if set up, there is the assumption that the blog resides on a non-secired session – using http: as protocol. So I missed all stylesheets. This is not aproblem since eveey link you need is available. So the configuration page does come up when selected and the protocol can be changed from http to https, leaving avery other bits in place – and all of a sudden, the blog comes up formatted, colored and all.

A first impression: It’s way faster than 2.3, and it looks the database-engine is less heavily accessed, though on a first post, MySQL crashed again – just after the watchdog loop started again, so I had to wait almost 15 minutes…. But after that, the impression was strenghtened.
Time for exploring the benefits (and drawbacks).

CIFS installed
CIFS – or Common Internet File System – is usually known as Samba. This is the first official, HP-supported version that is to replace AdvancedServer on Itanium, but there is also an Alpha version – and I installed today. Since it comes in an PCSI installation kit, it’s just $ PRODUCT INSTALL SAMBA /DESTINATION=AXP083:[CIFS'] (Warned before, there is no default! You’d better specify).

There are some post-installation tasks but these are well described.
The installation procedures leave SAMBA$ROOT defined (/SYSTEM) so that part of your work is done; but you still have to startup the environment, and define your commands. Once that is done, edit SAMBA$ROOT:[LIB]SMB.CONF, and you’re done.
Well, not exactly. Usually, that would be enough, but if you run Samba in a cluster environment, there is a little twist: create a file SAMBA$ROOT:[LIB]SMB.CONF_<Nodename> and include it in SMB.CONF.
I have all definitions in the node-specific file at the moment but I assume it just needs the node-specific issues.
One thing to consider it what type of authetication you want. The documentation does specify some hints, default is “user” but that bypasses the SYUAF authentication. I tried VMS and it sems to work – but I found no reference to this in the docs on the Samba site, not in the ones supplied by HP.
But since the recommendation is to use either LDAP or TDB (Trivial DataBase)-based authrntication, and I don’t have done anything on LDAP, TDB was the way to go. An empty file does already exist and so I could add my own account.
Restarted Samba – and behold: the basics work. Not exactly what I wanted, but the start is done. Logging on a kind of public directory with all user directories – but my own was there – adn accesable.

Next is promoting Diana to be a PDC. Again, the documentatio doesn’t tell how – but an Interex VMS-SIG session on the subject supplies the requirements. These have been added, and one full 18Gb disk as an extra, public share. Restarted Samba (Sinf the configuration file was changed) and it all worked somewhat nicer – login directory is really SYS$LOGIN and nothing else, and the public share was available as well. I now could save some files off my PC – onto a share. It took about an hour to copy allmost 2GB…

Now it’s a PDC, I could try to add at least Aphrodite as a member into the domain. It starts alright, but at end the system (what system?) askes for a username, passowrd and domain that is privileged to add a node.
Whatever I add, it fails. Either the format is incorrect, or the user is unknown, or the password is incorrect…. To have at lest one user available for this action, I added myself to be Administartor – a built-in group. To no avail – it still is impossible to add a node to the domain.

I must have missed something here.

To be continued

System Updates
A few weeks ago the next update was released by HP – one that consolidates a lot of updates since, well, some time ago. There were no patches excluded – so it would conclude all that I missed. In the following days there wer 4 or 5 product patches, so I downloaded them all and installed the whole bunch. A number of them required a reboot, so I pusched them all onto Diana and did one reboot. Probably not the best solution but the fastest. A full backup of the system disk takes at least 30 minutes, and each reboot takes another 5….When it would go wrong, I would be able to revert – of course I made a backup of the system disk in advance 🙂
On-line, /IGNORE=INTERLOCK – which is no problem at all since the files that are used on each cluster memner are taken off the system disk.

Testing after reboot: the blogs all work, MySQL is working properly. The VMS wiki works as well, so that part is set up correctly as well. I added a small chapter on patches.

Well – there is one problem: MySQL. Again.
I did the updates using CSWB directly on the VMS box. This has a huge memroy footprint, and so does MySQL, and MoinMoin, Adding up, it’s been too much for MySQL so it crashed, AFTER the update was done.
Well, it’s that way.

Getting on
with the VMS wiki. The start page is there, the links pages has been created and I did the Installation page. Looking at the system performance, most resources used by Python are – for this wiki anyway – just CPU and paging – but the memory footprint seems less, much less than PHP’s – or perhaps I should say: WordPress. At least, MoinMoin doesn’t need a database – and part of the big memory requirement wil, no doubt, be MySQL.

I’ve been working on the webpage program as well these days but it’s not progressing very fast. I ran into a number of problems I have to figure out first. I’m thinking of reversing some decisions since I now run into rather serious trouble in placing data properly.
Forgot to adhere to the KISS idea: Keep It Stupidly Simple. So it’s back to the drawing board…

VMS wiki running
Creating a new wiki – now for public access – is easy: just run a procedure stating where it should reside, change protection on all files in one location (I guess JFP is going to repair that twist also), create the mapping in the server configuration, and you’re done.
It turned out to be that easy indeed – after spending a considerable amount of time to figure out WHY it didn’t work.

The new wiki is located on a directory: webdisk:[wikis.vmswiki], and I already mapped it in the webserver configuration, but time after time, I got:

directory or file does not exist: /moin_wiki_root/vms/moinwsgi

and I could prove it did. moin_wiki_root = webdisk:[wikis.vmswiki] and it does contain the file I needed.

One thing the procedure does is running @logicals.com. and it tuned out this file contains the twist casuing the failure: at one point, the current directory (found to be Drive:[Web.wikis.vmswiki.vms] is adjusted by removing all [000000 and .VMS parts, and the result is used to define logical moin_wiki_root: drive:[wikiswiki] – the “.VMS” simply trimmed away!
No wonder it failed.
I renamed the directory to prevent this removal, and adjusted the mapping accordingly, and behold: problem solved: it works

Next is moving all static pages that way, and allow adjustments, and add a link in the front page. There is still a lot to learn on this: multi-linguality, security…
A first access will take a lot of time – and, for what I’ve seen so far, a considerable amount of CPU power, but after that, it’s pretty fast. The memroy footprint is less than PHP/MySQL, is seems, CPU being the highest used resource. Well, there is enough power available in that matter.