21-Jun-2008

Holidays
went over without real problems than the known issues – mainly lots of spam quarantained and rejected (apart from blocked). Just one file needed to be removed to have some slipped messages no longer appearing. Well, just a few, but at Process, an update contained a file that shouldn’t have been released in the first place and that one would cause these messages slip the maze. We’ll see.
Of course, MySQL did require some restarts these weeks, but far less than before. It seems that tyhe Bootcamp2008 blog is now less frequently accesses (haing it blocked in ROTOTS.TXT now) and load of the server has lowered to previous levels.

Next bug job is to have all pictures of the Beijing and Boston trips, and all holiday pictures fit for publication. beijing is mostly done, if I recall correctly, but memory fades after 4 weeks of heavy mental work 🙂

Rebuilding MySQL extension
I found the original remarks on rebuilding the MysQL driver for PHP on JFP’s site, and followed exaxtly the same steps as I did before. That is: I had to add a few extra logicals and have some renamed, and found just one (!) file: flock.h seems missing:

I1 18511 #ifdef VMS
I1 18512 /*
I1 18513 ** If we're using flock, then include the flock header
I1 18514 */
I1 18515 #if HAVE_FLOCK
I1 18516 #include "flock.h"
.1
%CC-F-NOINCLFILEF, (1) Cannot find file "flock.h" specified in #include directctive.

Scrolling up the listing, this is in zend_api.H. After having flock.h found in the [portlib] subdirectory – and adding one more logical to acess it easily – I got rid of this error, but stumbled into the next:

I2 21108 #pragma __standard
I2 21109 #endif /* __PWD_LOADED */
I1 21110 #include
.1
%CC-F-NOINCLFILEF, (1) Cannot find file
specified in #include directive.

More of a problem, since this seems to be within a DECC library routine, according the page header:

Source Listing 5-JUN-2008 22:21:36 HP C V7.3-009-48GBT Page 387
21-NOV-2007 09:42:33 SYS$COMMON:[SYSLIB]DECC$RTLDEF.TLB;1(PWD)

There must be some setting that needs to be changed, but which one?

There is some rememberance of thius sys/param.h system file – but I cannot recall what….

What’s more important: The orginal MySQL engine I built was against the base version pf MOD_PHP. The latest version is ECO-2, but what I found most recent as sourcekit, was ECO-1. It may make the difference. Just checked, and it seems the latest version is now available – including PHP ECO-2. So I got that one to be installed, but it might well be it won’t solve this issue…

01-Jun-2008

May statistics

PMAS statistics for May
Total messages    : 3690 = 100.0 o/o
DNS Blacklisted   : 2500 =  67.7 o/o (Files: 32)
Relay attempts    :   45 =   1.2 o/o (Files: 25)
Processed by PMAS : 1145 =  31.0 o/o (Files: 32)
        Discarded :  371 =  32.4 o/o (processed),  10.0 o/o (all)
     Quarantained :  442 =  38.6 o/o (processed),  11.9 o/o (all)
        Delivered :  332 =  28.9 o/o (processed),   8.9 o/o (all)

Again, about 70 is skipped right away, and only 9% is delivered. 16 messages slipped the check and made it to SMTP that filtered the out.
I had a day and a half that there seemed to be no blocked, discarded or quarantained messages, for some reason PMAs just stopped noting them. That is: there must have been such messages but these were simply not added to the database, until the configuration was reloaded (due to enabling debug…), after which they did appear again. Even after disabling debug.
So there may heve been a bit more blocked, discarded or quarantained….

MySQL has had it’s troubles this month, probably by the attention drawn by the Bootcamp 2008 blog:

MYSQL Restarted 2008-05-05 10:05:00.45
MYSQL Restarted 2008-05-11 17:05:00.75
MYSQL Restarted 2008-05-19 12:05:00.34
MYSQL Restarted 2008-05-21 02:50:00.52
MYSQL Restarted 2008-05-23 02:50:00.15
MYSQL Restarted 2008-05-23 14:20:00.52
MYSQL Restarted 2008-05-24 12:20:00.27
MYSQL Restarted 2008-05-27 06:05:00.65
MYSQL Restarted 2008-05-27 06:50:00.36
MYSQL Restarted 2008-05-27 08:20:00.53
MYSQL Restarted 2008-05-27 09:05:00.51
MYSQL Restarted 2008-05-27 13:35:00.45
MYSQL Restarted 2008-05-27 14:20:00.84
MYSQL Restarted 2008-05-28 02:35:00.31
MYSQL Restarted 2008-05-28 11:50:00.51
MYSQL Restarted 2008-05-28 20:20:00.13

As can be seen, the majority of crashes occurred during and immediately after the bootcamp (Ian Miller set a link on OpenVMS.org). All load on Diana has gone up and remains at that level up to this moment. Following are the different graphs by Hyperspy; the gap in the beginning of the month has to do with a wrong definition of the logical pointing to the Hyperspy root directory – due to a rerun of the script starting it all. You can see the load increases dramatically on May 14th and stays high from that time on:

CPU load:
CPU load May
Memory usage:
Memory usage in May
Pagefaults:
Pagefaultrate in May
Processes:
Number of processes
Buffered IO:
Buffered IO in May
Direct IO
Direct IO in May
(The big hump on May 27th coincides with MySQL problems thaty day, and it may have to do with uploading a lot of data)

As for FTP, only thee were found in oeraror logs this month (searched from the saved files):

$ sea operator*.txt FTP/window=(1,5)

******************************
WEB_DISK2:[private.operator]OPERATOR_done1664.txt;1

%%%%%%%%%%%  OPCOM  14-MAY-2008 15:53:27.93  %%%%%%%%%%%
Message from user TCPIP$FTP on DIANA
        User Name:      anonymous
        Source:         server213-171-222-110.live-servers.net
        Status:         NOPRIV -- File access violation
        Object:         WEB_DISK2:[public.anonymous.080514155324p]

******************************
WEB_DISK2:[private.operator]OPERATOR_done1669.txt;1

%%%%%%%%%%%  OPCOM  19-MAY-2008 20:06:48.38  %%%%%%%%%%%
Message from user TCPIP$FTP on DIANA
        User Name:      anonymous
        Source:         59.22.140.17
        Status:         NOPRIV -- File access violation
        Object:         WEB_DISK2:[public.anonymous.test]

******************************
WEB_DISK2:[private.operator]OPERATOR_done1673.txt;1

%%%%%%%%%%%  OPCOM  23-MAY-2008 10:39:07.93  %%%%%%%%%%%
Message from user TCPIP$FTP on DIANA
        User Name:      anonymous
        Source:         88-117-50-55.adsl.highway.telekom.at
        Status:         NOPRIV -- File access violation
        Object:         WEB_DISK2:[public.anonymous.080523114246p]

There were a few more but these were ‘known accesses’ – getting data that was too much to send by e-mail.

All logs have been archived.