23-Jul-2015

Results of more memory
The main question is: Did this memory expansion any good? It should be visible in the output of the performance tools.
First: Hyperspy++ (that comes with WASD): It does show memory usage is much lower, but consistent with the prevous size (about a quarter) and paging hasn’t changed much:
2015-07-23_08-24-39
T4 output shows similar freelist size but the modified page list has increased:
2015-07-23_08-30-00
and here as well, paging statistics haven’t changed much:
2015-07-23_08-30-37
which is to be expected, since memory utilization is pretty much the same:
2015-07-23_08-33-09
Seems odd, but is to be expected: Since I still have to do a reconfiguration of the paging parameters – that’s what AUTOGEN is all about. It is planned to be done somewhere next week.

But on the subjective side, there are improvements: it is obvious that the increased size of the modified page list allows faster execution of PHP code. No ACCVIO or stack overflows so far; just a few:

%HTTPD-W-NOTICED, 22-JUL-2015 19:52:40, CGI:1997, not a strict CGI response
-NOTICED-I-SERVICE, http://www.grootersnet.nl:80
-NOTICED-I-CLIENT, 82.161.236.244
-NOTICED-I-URI, GET (100 bytes) /sysblog/wp-admin/admin-ajax.php?action=dashboard-widgets&widget=dashboard_primary&pagenow=dashboard
-NOTICED-I-SCRIPT, /sysblog/wp-admin/admin-ajax.php sysblog:[wp-admin]admin-ajax.php (phpwasd:) SYSBLOG:[WP-ADMIN]admin-ajax.php
-NOTICED-I-CGI, 504850205761726E696E673A20204D6F64756C652027646F (2048 bytes) PHP Warning: Module 'dom' already loaded in Unknown on line 0.
-NOTICED-I-RXTX, err:0/0 raw:1154/0 net:1154/0

yesterday from the workstation, and one today. But it all seems to work without problems.

SCSI hardware is already present
When changing the memory, I also realized there are two SCSI cards in the box: One serves the internal disk (that holds the page- and swap files), and the other connects to the external storage unit. The second is surely a KZPBA, the first one probably is as well, it too has an 68-pin outlet at the back. So there is no real need to install another KZPBA – it’s all there.
The one thing to find out is to see whether terminators are there, or not. Access is not that easy so I didn’t check last night. So there well be another (short) outage, probably in a weekend.

Hot, hot, hot
Tonight everything was stalled. Not even the console triggered a reaction. Normally, if the screen has gone blank, there are two leds flashing on the keyboard but even these were unlit. The reset button that would normally invoke ^P, had no response. So I stopped the machine the hard way (powered it off) and powered it o again, and there I found the system temperature was 44 degrees. So the machine was probably halted due to high temps. Restarted it (and found a bug in WASD?), cooled it somewhat by pushing air through the vents in front. Now hoping it won’t get that hot. (Yes, I know. I need airco in here. As soon as I have the bucks, I’ll get one)

22-Jul-2015

Server errors
Sometimes it seems impossible to acess the blogs, and the WATCH output shows 500 errors – with no particular text. The server log however does.
Fairly often there is:

%HTTPD-W-NOTICED, 22-JUL-2015 08:21:54, CGI:1997, not a strict CGI response
-NOTICED-I-SERVICE, http://www.grootersnet.nl:80
-NOTICED-I-CLIENT, 68.180.229.122
-NOTICED-I-URI, GET (20 bytes) /sysblog/?m=20110727
-NOTICED-I-SCRIPT, /sysblog/index.php sysblog:[000000]index.php (phpwasd:) SYSBLOG:[000000]index.php
-NOTICED-I-CGI, 504850205761726E696E673A20204D6F64756C652027646F (2048 bytes) PHP Warning: Module 'dom' already loaded in Unknown on line 0.
-NOTICED-I-RXTX, err:0/0 raw:196/0 net:196/0

and once in a while:

%HTTPD-W-NOTICED, 15-JUL-2015 00:12:17, CGI:2107, not a strict CGI response
-NOTICED-I-SERVICE, http://www.grootersnet.nl:80
-NOTICED-I-CLIENT, 89.248.167.137
-NOTICED-I-URI, POST (19 bytes) /sysblog/xmlrpc.php
-NOTICED-I-SCRIPT, /sysblog/xmlrpc.php sysblog:[000000]xmlrpc.php (phpwasd:) SYSBLOG:[000000]xmlrpc.php
-NOTICED-I-CGI, 2553595354454D2D462D53544B4F56462C20737461636B20 (66 bytes) %SYSTEM-F-STKOVF, stack overflow, PC=FFFFFFFF8083B42C, PS=0000001B
-NOTICED-I-RXTX, err:0/0 raw:462/0 net:462/0
or

%HTTPD-W-NOTICED, 22-JUL-2015 08:51:37, CGI:2107, not a strict CGI response
-NOTICED-I-SERVICE, http://www.grootersnet.nl:80
-NOTICED-I-CLIENT, 192.3.154.170
-NOTICED-I-URI, POST (19 bytes) /sysblog/xmlrpc.php
-NOTICED-I-SCRIPT, /sysblog/xmlrpc.php sysblog:[000000]xmlrpc.php (phpwasd:) SYSBLOG:[000000]xmlrpc.php
-NOTICED-I-CGI, 2553595354454D2D462D41434356494F2C20616363657373 (118 bytes) %SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000003B22676E, PC=FFFFFFFF8083B42C, PS=0000001B

that occurs in different places within the PHP engine.

I restarted the server and for now, the blogs can be accessed again.
The problems have been signaled but I didn’t get an answer yet. Stack overflow and Access violations may be related to the lack of memory – which is to be handled shortly.

Update
On posting an update, I got a complete stackdump on screen (without formatting, of course). This may help[ to solve the issue:


%SYSTEM-F-ACCVIO, access violation, reason mask=00, virtual address=000000002D70772F, PC=00000000003F6870, PS=0000001B%TRACE-F-TRACEBACK, symbolic stack dump follows image module routine line rel PC abs PC PHPSHR 0 00000000003A4870 00000000003F6870 PHPSHR 0 00000000003A9E74 00000000003FBE74 PHPSHR 0 00000000003E3894 0000000000435894 PHPSHR 0 00000000003A43B8 00000000003F63B8 PHPSHR 0 00000000003B58A4 00000000004078A4 PHPSHR 0 0000000000370B8C 00000000003C2B8C PHP52-13WASD PHPWASD ProcessRequest 47240 00000000000020F8 00000000000320F8 PHP52-13WASD PHPWASD main 46824 0000000000000728 0000000000030728 PHP52-13WASD PHPWASD __main 46717 000000000000006C 000000000003006C PTHREAD$RTL 0 0000000000057718 FFFFFFFF80A77718 PTHREAD$RTL 0 0000000000030444 FFFFFFFF80A50444 0 FFFFFFFF8037FC44 FFFFFFFF8037FC44%TRACE-I-END, end of TRACE stack dumpPHP Warning: Module 'dom' already loaded in Unknown on line 0 PHP Warning: Module 'session' already loaded in Unknown on line 0 PHP Warning: Module 'SimpleXML' already loaded in Unknown on line 0 PHP Warning: require(/sysblog/000000/wp-includes/author-template.php) [function.require]: failed to open stream: not enough core in /sysblog/000000/wp-settings.php on line 126 PHP Fatal error: require() [function.require]: Failed opening required '/sysblog/000000/wp-includes/author-template.php' (include_path='.:/php_root/000000') in /sysblog/000000/wp-settings.php on line 126 PHP Fatal error: Maximum execution time of 60 seconds exceeded in Unknown on line 0 PHP Fatal error: Cannot redeclare redirect_post() (previously declared in ðÅCblog/wp-admin/SYSBLOG:[WP-ADMIN]post.php:41) in /sysblog/wp-admin/SYSBLOG:[WP-ADMIN]post.php on line 83

Update 2
Didn’t want to wait. The 2Gb memory has arrived and I decided to install it right away.

$ sho mem
System Memory Resources on 22-JUL-2015 20:20:41.50


Physical Memory Usage (pages):     Total        Free      In Use    Modified
Main Memory (2.00GB)              262144      168149       87963        6032

Far better figures than before!
All in all it took less than half an hour. After restart, it became clear that the startup needed to be adapted as well, wasn’t done yet, so it took several more minutes to get it all up and running again. There were a few issues, mainly logicals that needed to be redefined. But once that was done, all went well.
It seems to be quite an improvement: the blogs loads faster (once the runtime environment is loaded and initialized) and since there is much more memory, my guess is that paging – especially hard faults – will be less. But that will show within a few days.
To get things even better, the system has to run for some time after which I could do an autogen to re-configure the system – memory in particular.
Time will also tell whether ACCVIO’s and stack overflows are now a thing of the past.

10-Jul-2015

Memory shipped
Just got a mail that the memory I ordered has been shipped, estimated delivery date is 13-Jul-2015 – so next week. Though I planned to add a SCSI card – since the DS10 has no external SCSI connector, a second card is needed – I will do later: I think it better to rearrange the site. Plus that I may have a spare somewhere: There still is some hardware stored and a SCI card may be in there somewhere.
Windows 10 preview
The previous workstation (Helena) has been set up to preview Windows 10, and, for as far I have been working on it, well, it does it’s job. No surprises, the software I normally use works, there was one nasty thing that I need to check. If there is something wrong with a site’s certificate and you still want to access the site (because you KNOW it is not a bad one – like mine :) ) you can not.
This is caused by the fact that I use one (self-signed) certificate for all my secured web-entrances, and the site’s names differ from the one in the certificate. And it needs a renewal as well…
Currently, the latest update is being applied (10160) and if all goes well, I see no reason why not upgrading some of the systems..

07-Jul-2015

Ordered more memory
The updates of PHP and WordPress seem to require more memory: the blog loads slower, and paging has increased. This was expected – but not in this amount. So I ordered more memory to fill the system to the max. It will take some time to get it, and installing it means the system won’t be available for some time. I take the opportunity to install a SCSI card as well so I can hook up the disks directly to the machine, and get rid of the HSZ50 unit: the battery is gone so caching is disabled and access is now limited to 5Mb/sec; since I have only one controller board, it is of no use for clustering: I tried but it simply doesn’t work with the Workstations: these crash dramatically. Or does someone have left-over controller board, cache and battery units?

01-Jul-2105

Maintenance
The log of the maintenance job shows no real surprises:
PMAS statistics for June
Total messages    :   1552 = 100.0 o/o
DNS Blacklisted   :      0 =    .0 o/o (Files:  0)
Relay attempts    :    105 =   6.7 o/o (Files: 30)
Accepted by PMAS  :   1447 =  93.2 o/o (Files: 30)
  Handled by explicit rule
         Rejected :    725 =  50.1 o/o (processed),  46.7 o/o (all)
         Accepted :    323 =  22.3 o/o (processed),  20.8 o/o (all)
  Handled by content
        Discarded :    180 =  12.4 o/o (processed),  11.5 o/o (all)
     Quarantained :    183 =  12.6 o/o (processed),  11.7 o/o (all)
        Delivered :     36 =   2.4 o/o (processed),   2.3 o/o (all)

and in the logfiles, it shows just on 7th, 16th and 26th of June more relay attempts than usual; the files are 8 blocks (4K) in size, so not that big.

However, I have AkisMet running on the blog and that shows over 800 spam messages for June – just these few days, and almost 800 for today alone; however, it may be related to the installation of the new version of the software installed tonight.

Fot the blog: I downloaded a number of themes that I like on first glance, and installed them. It’s now a matter of determining which one will be applied. It may take a few switches in the coming weeks. For the time being, I customized the twentyFifteen theme that comes with WordPress. Next I encountered a number of issues with the editor screen when enabling the excerpt form: but this is admin only so I don’t bother too much – for now.

29-Jun-2015

Blog updated
FINALLY.
After I added extra definition to the configuration file, to block access other than my own; changed php.ini to allow the upgrade script to run for more than 30- seconds (gave it 10 minutes to execute), disabled the plugins and reverse to the default theme – recommended by WordPress – and applied the upgrade script, I have now the SYSMGR blog running on WordPress 4.2.2. There was one final thing to do to allow access: Change the basic theme because ” default” does not exist in this environment. Changed it to twenty-fifteen, for the time being, because the previously used theme is not compatible with this version of WordPress. I downloaded a few that seem to meet my requirements and of which the look and feel seem nice, but I still have to install them and than decide which one will be the next one.
Both plugins:L Askimet and Stats, need to be updates as well but that will be done later as well.

The Trips, Tracks and Travels blog will be upgraded soon – the same way.

05-Jun-2015

Latest news on PHP
Got two messages from Mark:
1. PHPWASD.EXE in the kit is no good.
2. He reviewed all kits and rebuilt them when needed – there should be no more mismatches any more.
To be tested tonight, or during the weekend.

Evening update
I downloaded the updated kit from Mark’s site and installed it. First, it didn’t work since PHPWASD: couldn’t access PHPSHR, and later, it seemed that the MySQL extension was missing. The reason was just a matter of file protections’, I had to change the n all, and edit php.ini to enable the extensions (PHP_MysQL.EXE in particular -PHP_MySQLI.EXE alone wasn’t enough for this version of WordPress) but now it works, though it looks to be a bit slower. However, I should NOW be able to update to WP 4.2.2. Fingers crossed :)
Evening update 2
Updating WP means: revert to default theme and disable plugins – after creating a backup (so if it fails, it’s easy to revert to the stae before the update)/ No it should be a matter of changing the logical (the second location should be where the new version of WordPress is stored, in my case WP42:), than restart the webserver (to remove any running PHPWASD images), and start the blog. It does start with the mentioning that WP version has been changed and that the database needs an update, It starts – and stops without a notice, somewhere in step one of the process.
So I had to revert to the state prior to the update, by simply restoring the database, change the logical back to the original version and restart WASD.
Now it’s a matter of finding out what is going on….
It may be related to the resource limits in PHP.INI:

;;;;;;;;;;;;;;;;;;;
; Resource Limits ;
;;;;;;;;;;;;;;;;;;;

max_execution_time = 30 ; Maximum execution time of each script, in seconds
max_input_time = 60 ; Maximum amount of time each script may spend parsing request data
;max_input_nesting_level = 64 ; Maximum input variable nesting level
memory_limit = 128M ; Maximum amount of memory a script may consume (128MB)

that may be too low, keeping in mind that this system is limited in resources…

Maintenance
It has been a few days ago that the maintenance jib has run. No surprises:
PMAS statistics for May
Total messages    :   1026 = 100.0 o/o
DNS Blacklisted   :      0 =    .0 o/o (Files:  0)
Relay attempts    :     88 =   8.5 o/o (Files: 31)
Accepted by PMAS  :    938 =  91.4 o/o (Files: 31)
  Handled by explicit rule
         Rejected :    387 =  41.2 o/o (processed),  37.7 o/o (all)
         Accepted :    197 =  21.0 o/o (processed),  19.2 o/o (all)
  Handled by content
        Discarded :    152 =  16.2 o/o (processed),  14.8 o/o (all)
     Quarantained :    185 =  19.7 o/o (processed),  18.0 o/o (all)
        Delivered :     17 =   1.8 o/o (processed),   1.6 o/o (all)

The number of relay attempts has been minimal this month: No file exceeded 10 blocks; there have been some, of course, but only three files were large enough (over 2K in size) to be examined: mail was sent from addresses 173.254.223.72 , |185.60.229.89 and 5.79.68.231; faked senders, no doubt, since From: was either empty, a (non-existing) user from grootersnet.nl (and I do know my users, and my address :)) of admin from a site that is not related to the address it is sent from.

04-Jun-2015

More tests on PHP upgrade
I got a reminder of Mark Berryman, to re-download the kits and install them.
I started with PHP 5.2.13, the one I testes two and a half year ago or do and found in not working. I still had a PHPWASD executable of that one – separate – because it became clear that there was a problem with that file – or with PHPSHR.EXE.
But this one wasn’t working either, I got “Ident mismatch with shareable image” again, so the kit wasn’t right. I tried again with the separate image, which appeared tp be from a later date, but the same happened.. Analyzing the images, it became clear that these two images were the same. Also it was proven there indeed was a mismatch between the programs and the shareable image. So I signaled this to Mark.

31-May-2015

More on PHP and WordPress upgrades
After yesterday’s mistake, I had to revert the activity from PHP 5.4 back to 5.2 – it is just execution of a script, without a problem.
First thing today was re-trying the blog with PHP 5.4. Setting this is no problem, PHP_INFO.PHP works – but none of the blogs do because of deprecated and non-strict syntax – but these wouldn’t be the biggest problem since I must have got them yesterday as well. To get them out of the way anyway, I redirected this logging to a file, but that didn’t solve the problem: the last error typically is:
PHP Parse error: syntax error, unexpected end of file in /sysblog/000000/wp-includes/post-template.php on line 734
The code does indeed miss the training end-tag:
<?php
else :
echo "<ul class='post-revisions'>n";
echo $rows;
echo "";
endif;

[EndOfFile]

So no ?> tag on line 734 (on EndOfFile line) which is in fact simply completely valid.

Yesterday there wasn’t a problem with this, so why there is today?? This means I have to revert to 5.2….
(I now know why. I didn’t close down PHP sessions so the current version still was 5.2, but unseen)

Second, I wanted to create a new blog using WordPress 4.2.2. I updated the file containing the database credentials, started the blog which introduced the 5-minute install. No problem – except that is stalls on step 2, the page doesn’t show anything except for a white rectangle, there should be a first-level header but it seems the page isn’t complete. No error message, nothing.
So I dropped the database (the right one) and recreated it (it must exist) but now I cannot connect to the database – as shown in the WATCH output….
Even on 5.2, it is a problem. Not really strange if the mapping is false….Once that was settled, I can get it running on 5.2 to start with. But it seems it doesn’t use wp-config.php, so take a look there: matter of protection: Should be readable. Once that is settled for all files, installation fails on the glob function – which is missing in this PHP version – now I remember why I wanted this upgrade….
However, the database does exist and tables have been created; some have been filled as well.
But now this is settled, I could retry re-installing the 4.2 version using PHP 5.4. But that fails – again, where PHP_INFO just works.
Tried it with 5.3, but that fails as well, but for no apparent reason: just a 500-error….And 5.2.13 (re-installed) runs into
Ident mismatch…
So I seem to be stuck with this version.
I could however try upgrading the blogs to 4.2.2 directly: Change the logical to use WP42 in stead of WP265.
But no. I need to switch to all defaults before and I forgot. So returned to the previous setting – with a typo caused NoService to appear! Even after repair, where Trips, Tracks and Travels have no problem…..
(As it turned out – after having saved last entries using mySql and save them into a file and restoring yesterday’s backup – it was just the cache of InternetExlorer that caused the problems….)

Give it another try: Update requires all specifics switched off and running the update script as administrator. Before doing anything, I backup the database, then switched to WP4.2 and ran the update script.It showed that a database update was required as well, I started it but that one – as wll – stoped on a fatal error: the glob() function being undefined.
Duh.
This is still PHP 5.2.6….
Restored it all: WP version and database. Wait until I get WP up and running using PHP 5.4 (or higher: I have downloaded 5.5 as well but I think it will have the very same issues as 5.4).

Informed Mark Berryman, perhaps he has something to solve the problems….

30-May-2015

PHP Upgrade
Finally, I got some time to upgrade PHP and WordPress.
First things first: Do PHP first; I already tried 5.2.13 but ran into trouble with that. The next versions were 5.3 and 5.4, and I decided to go for 5.4. The procedure to establish this version needed a slight adaptation but once done, all logicals were set properly. But running a PHP-script by the browser always ended in an error: Ident mismatch with shareable image – being PHPWASDSHR, referring to PHPSHR.EXE. I contacted Mark Berryman, who suggested to download the kit again, install it and see if that worked.
It did.
PHP_INFO nicely showed the information, and even the SYSMGR blog was shown with no problem at all. So I could do some blogging – and there it showed there were still some issues with starting the page to add a new post (new-post.php) or, from another session, managing posts (edit.php), but on reload, it all seemed to work nicely – even without restarting the server. Since I planned to update WordPress as well, I didn’t bother too much.
Next step was the installation of WordPress 4.2.2; I use the zip-container since my version of VMStar extracts the files in all uppercase letters; both however won’t accept dots in a name, except as delimiter to filetype, and replaces them by underscores. Luckily, WordPress doesn’t use underscore in filenames, so it’s an easy step to change them back to dots, using a small script (I found the original after this!). Next step was creating a new blog, but for some reason, it stopped in step 2. So I decided to retry after cleaning up.
Here I made a mistake. I dropped the WordPress database -which contains all blogs….

The last thing I could do was restore it from last nights backup and continue working tomorrow…

DNS issues on Helena
Apart from this: the Windows 8.1 Pro workstation suddenly failed to access the VMS box (Diana) by name. This effected all protocols: POP, SMTP, HTTP, TELNET, SSH…. Ping on the name (requiring translations) didn’t work either. As it turned out: The machine considered itself member of domain grootersnet.local – I set that up in the zone-file of BIND on Diana but never implemented it; however, Helena found should be within this domain, probably because the .local extension seems to be some sort of default (I’ve seen it else where and I recall having similar issues in the past). Bow I removed it from the BIND.conf file on Diana, including the files that describe it, and restarted TCPIP$BIND. After that, translations went fine again.