Month: July 2008

More installations
On the Personal Alpha – normally named “Persephone”, or Prese, for DECNet – I installed quite a bunch of software: Languages to start with, and that caused a new systemdisk container to be made of 4 Gb – the max – to hold all software. Most could be done using the program that resides on the CD, just Cobol required a different approach because the cdmenu program didn’t find an installable product – so this one needed to be done by hand. But at end, all major languages are installed but need to be started. I also installed FMS and DECForms, and would like to have ACMS as well – but that requires RdB, CDD and a working DECNet connection.
DecNet is some other thing I would like to have running, and on the PA environment it was fairly easy – but to get it to work, I found I needed Diana to be configured as an end-node – it didn’t work as an L1 router. Problem however is, how to restart DECNet without interrupting the rest of the network traffic? I tried long ago and cause the system to crash – requiring a reboot to get it fixed…Well, wait for a next boot.
It came earlier than anticipated.
Retry the cluster
I retried the shared-SCSI cluster once more, not to get it running but to get output to be discussed with Colin Butcher (of XDelta) who offered me a hand in this installation during this year’s bootcamp. He’s too busy getting a big cluster up-and-running at the moment and I’m not in a hurry so he can take his time and answer when he thinks he can.
In this process, I started the AlphaServer 400 from it’s own internal disk, but all licenses were expired. This was known and easy to get by, the procedure is available – on the disk containg the clusterwide files, which is connected on shared SCSI. I mounted it – but this was a mistake: It caused Diana to hang and I had to reset it the hard way by pushing the RESET button….
DECNet was now working: I could access Diana from the PA using DECNet, by address and by name. And the other way round as well.
As a side effect, I could also check what was properly added in the startup yesterday, and what was missing. RdB, for example, wasn’t started because the procedure taking care of all RdB stuff wasn’t invoked. My change to the home page text – that logical – was properly done, I got the default’ Rebooted’ homepagetext – so that’s good. The wiki isn’t displayed properly – again a missing logical that I should have added. But RdB and related software were started nicely. At least, at first glance.
Somewhat more on RdB
But there was still something with RdB: the TCP service RDBSERVER didn’t start, because the user wasn’t known, something must have been wrong during installation. It wasn’t solved easily – I could imagine the username would have been TDB$REMOTE72 – but that didn’t work either. So I stopped the database monitor, removed the service and redid the installation. In the IVP the database – and the service – were now started as it should.
Although it seems the CDD is installed – CDD logicals have been defined, and the package has started – manually, it’s not yet in the startup – the command I would expect: CDO is not known. Well, it may be something else, this is to be looked at as well.

Installing RdB
One thing I wanted to do for a long time is installing RdB – the only really VMS-based relational database. Once one of those Digitlal products that were too far ahead of their time, and lacked the markering it deserved, now still in active development at Oracle.
Installtion should be easy, but it turned out that I had a few things messed up slightly: since all clusterwide files are placed in a separate location and therefore are reeferenced by logicals, these should all be correct. As found by this installtion, one wasn’t: VMS$PASSWORD_HISTORY had a typo and referenced a non-existing device. By that, SET PASSWORD/GENERATE as used by the installation procedure didn’t work and so the installation aborted. The documentation warned that if the password should not be locked – but here, the whole account didn’t exist, but this mismatch wasn’t mentioned. Oracle’s Metalink RdB forum proved to be helpfiul, when the problem was solved, installation went flawless. The RdB monitor and processed are now started, taking very little resources.
A second program to be installed is SQLServer – the SQL frontend, allowing external access to the database by SQL*Net – like an ordinary Oracle database. No problem here either.
A third product I’d like to have installed that relies on RdB is the CDD – Common Data Dictionary, but installtion here fails because the distributed transaction manager isn’t running – because DECNet has not been configured – yet.
So some work to be done.

Patches and new software installed
and some more bits of new software: A new Python environment (that has been around for some time) has also been enabled, the database disk container, just holding the MySQL log files has been shrinked to a quarter of it’s size; Samba re-installed (but it still needs to be tested)… It meant adjustment of the startup procedures – minor, in fcat. A good moment since I did have to reboot after patch installation.
Immediately it became clear that the definition of the logical that holds the filename of the home page text was missing, and so the home page didn’t show up. It’s not in the startup at the moment, I’ll need to make a default to prevent it from happening.
Another thing that’s wrong: thw VMS wiki doesn’t start, caused by another reference to the previous Python environment. An easy change, and just execute the file. It’s just that the logo doesn’t show up, caused by another missing logical. It refers to the MoinMoin version, hardcoded, in the configuration file, so that’s to be changed as well – until the right logical is set (moin_static – no version number…). The other porblem is I’ll need to copy the images as well.
As it turned out: the VMS wiki mapping in embedded in the normal web configuration, not in an include file…. Once that was updated, all is well.
Rdb
Planned quite some time ago was an installtion of RdB , the real OpenVMS database. The system has been configured to cope with it’s requirements, and now it’s time to give it a try.
However, installation fails for unclear reasons on the creation of accounts. According the installation guide, the error would occur if the installing account would have a locked password (have the lockpwd flag set in UAF) but that is not the case. Got myself an Oracle Metalink account (thanks to the company) to ask. (Is it meaningfull that the message from the installation procedue mentioned the wrong chapter in the guide?)

June mail statistics
PMAS statistics for 06
Total messages    : 4257 = 100.0 o/o
DNS Blacklisted   : 2929 =  68.8 o/o (Files: 32)
Relay attempts    :  111 =   2.6 o/o (Files: 29)
Processed by PMAS : 1217 =  28.5 o/o (Files: 32)
        Discarded :  437 =  35.9 o/o (processed),  10.2 o/o (all)
     Quarantained :  514 =  42.2 o/o (processed),  12.0 o/o (all)
        Delivered :  266 =  21.8 o/o (processed),   6.2 o/o (all)

Apart from those that were either blocked, discarded or quarantained, 29 messages that had an invalid (non-existing) domain) were trapped by SMTP itself. This is a test not yet included in the product, but I can live with that: it’s less than 1 % of all messages. It seems it was just a burst lasting four days.
Also, massages that are obviously spam, did slip the mazes because one file was included in the last update that shouldn’t have been there – as Hunter kindly admitted. After having this file removed, that was over immediately – though a few got through but Process has been informed and has adjusted the filter.
Webserver and related
that means: including MySQL.
No real issues here this month. Since load has dropped somewhat, and the 2008 bootcamp blog now hidden for search engines, MySQL is keeping up nicely – was restart ‘only’ 11 times:

MYSQL Restarted 2008-06-02 13:50:00.32
MYSQL Restarted 2008-06-02 14:05:02.44
MYSQL Restarted 2008-06-03 23:05:00.56
MYSQL Restarted 2008-06-04 00:20:00.18
MYSQL Restarted 2008-06-04 06:35:00.72
MYSQL Restarted 2008-06-06 12:50:00.33
MYSQL Restarted 2008-06-06 16:20:00.65
MYSQL Restarted 2008-06-07 01:50:00.11
MYSQL Restarted 2008-06-07 22:50:00.41
MYSQL Restarted 2008-06-17 10:20:00.78
MYSQL Restarted 2008-06-18 13:50:00.20

I’ve been trying to build a new 4.1 engine because this (3.x) version is not capable of some queries in the admin panel – nothing critical but annoying nevertheless. However, it’s still failing to compile, and I have to update the PHP-part to ECO-2 (it’s there but needs to be installed).

I tried te reverse the settings in the configuration file bot for some reason, MySQL failed to stop, and after killing the server process the hard way, it refused to restart, because a log file seemed to be of the wrong value. So I used the previous (lower) setings but increased bytecount and pagefile quota by another 50%. Just to improve speed. After that, the server started nicely.And it surely looks faster now.
Why it failed to react on myadmin -u <username> -- password <password> shutdown, I don’t know. It should, after 4 or 5 times, wouldn’t it?

Other annoying errors from the past have no longer appeared so at that point, it looks better. Pity the MySQL engine is the old one…

For the rest – nothing is wrong, though I have to look into some issue in WASD that I asked around for, but it seems I’m the only one that encountered the problem. On the web-access front, it’s the usual stuff: w00tw00t, proxy-attempts (to www.yahoo.co.uk), packages that are probed, attempts to exploited vulnerabilities in pacakges installed on their default locations….

FTP
Now and than there are the ususal assumptions that this is a (standard) Windows or Linix site. Just consuming CPU and bandwith (shame on them) but no harm done otherwise.
And once in a while, Goolgebot comes along and tries again to access a directory that has been deleted eons ago. The last attempt however has been June 6th, so hopefully it learned.