11-Jul-2012

Overload
At about 22:00, I couldn’t get to the webmail agent – got a “503” error. Checked the server and found about 20 helper processes, many of them running PHPWASD (so accessing PHP code) and in LEF state. But the administration pages of WASD showed just one or two.
I tried restarting the server but that didn’t help.
I tried STOP/ID for each of them, no error but the processes didn’t go away.
The only way to get around it was to restart the machine. Next, I changed the throttle, it allowed for 20 processes and even more. So I lowered the number of processes for all blogs.

That should help.

05-Jul-2012

New hardware
My (about 5 years old Pentium-4 HT) workstation dubbed Aphrodite will get another role: It will replace the machine in the living room. That one is Pentium-4 as well – without HT, and quite problematic at times.
I obtained a state-of-the-art new box: ASUS P9X79 motherboards with Intel i7 processor @ 3.6Mhz, and 8 Gb of memory (expandable to 64Gb). 8 channel audio, as on my previous system. Transferred Video and disks, but not the DVD-drives: These are ATAPI and the new system has (e)SATA only, to I had to obtain that as well.
Suitable for the heavy stuff I intend to run on the beast: Running multiple Alpha emulators side by side, and processing sound, image and, perhaps, video. I could use Linux on the box, but the emulators I can use do either not run on Linux, or not in a way that I intend to use them. For sound- and image processing, I already have Windows-based software I can work with pretty well, and I would need to learn these Linux equivalents as well. In the future I may add Linux as an alternate OS but for the moment, I stick to Win7pro-64.
Some trouble: The front USB bus has a different connector that doesn’t fit anywhere on the motherboard ans yes: I do need them; there is no COM exit, it needs to be added, and the motherboard seems to have a broken DIMM slot so 4 Gb of memory isn’t fitted in it’s preferred position.
These I’ll have to address with the supplier.
But I had to install the OS from scratch since Win7Pro-32 didn’t boot on this box. But installing the OS didn’t work out as good as I expected: I had to clear the whole disk – including the partitions containing data – because the BIOS of the new box couldn’t handle them: this is EFI based….
It was no problem to move them to the other disk – I thought – using a DOS box and XCOPY the contents to a directory on the other disk. But once that was done, I couldn’t find the directories I created. No big deal. Pity – but I do have a backup, and there hasn’t been much changes after that anyway. Of would it be a disk I didn’t expect?
After shutdown and moving the machine to its fibal locatioen, it turned out that a boot after shutdown did almost always fail, and I had to do a repair from the installation disk – which invariably failed because “… the system to be repaired is incompatible…” . But when offered to reboot normally, there seemed to be nothing wrong. Might been caused because I installed in safe mode?
So in the end, I re-initiated the disk, and installed everything again from scratch – but now when booting from the DVD-drive. From that moment on, it all went smoothly. Getting drives and software from in Internet – no problem.
But in accessing any of my own sites, there was. None responded, but services and servers were up and running….
New IP address
First thing to be done is pinging the server by name:
ping www.grootersnet.nl
translated the server to be 85.223.43.24 – what I would expect because that’s what’s in DNS, being the outside address of the router.
But when accessing the router to seee what it says, the WAN connection is now on a different network, as well as the DNS-resolvers of the router. Contacted the ISP site (luckily there was no problem getting out!) and found there had been done some work that morning, and the connection had been down for a few minutes. So I called the help desk, and it was confirmed that the address had indeed changed. but there had been no information on this – which I would expect to be sent IN ADVANCE. Anyway, I had to contact the registrar of my domain to have the DNS references updated. That requires a signed document, which could be sent as an attachment in an email message.
Which I couldn’t use over the Internet….
But there are other addresses I could use: my provider’s, gmail, yahoo, hotmaill….So I created the letter, signed it, scanned it (using the new box – even with it’s problems) into a PDF file and mailed it. Next by the phone, it was handled within minutes, but it took some time before it would be expanded over the Internet.
This morning however, it still didn’t work: Although the new DNS-servers got them (the router configuration shows their addresses) the DNS servers inside the LAN didn’t. So I restarted BIND on the VMS box, and the router, but in some way or another it didn;t help. Looking into the LAN configuration of the router, I found the DNS-server in the LAN was the VMS box…Removed that: and now it’s all working again. But from elsewhere – mail, in particular – it may take some time: I didn’t mention that so that still refers to the old address..Will be changed today as well.
There has been one advantage: No spam either 🙂
Clean-up of DHCP and local DNS
Over the years,. systems have come and gone, and any new node in the LAN get’s an address by the DHCP server; and as long as the MAC address doesn’t change, that address will get the once supplied address. That will add these systems in the local DNS – and they’ll stay there.
But systems come and go, and the obsolete references are not deleted. So I took the possibility to remove all the old entries in both the DHCP and BIND databases.

02-Jul-2012

Regular maintenance
Just the usual.
PMAS statistics for June
Total messages    :   5380 = 100.0 o/o
DNS Blacklisted   :    979 =  18.1 o/o (Files: 30)
Relay attempts    :     48 =    .8 o/o (Files: 30)
Accepted by PMAS  :   4353 =  80.9 o/o (Files: 30)
 Handled by explicit rule
        Rejected :   3692 =  84.8 o/o (processed),  68.6 o/o (all)
        Accepted :    247 =   5.6 o/o (processed),   4.5 o/o (all)
 Handled by content
       Discarded :     65 =   1.4 o/o (processed),   1.2 o/o (all)
    Quarantained :    307 =   7.0 o/o (processed),   5.7 o/o (all)
       Delivered :     42 =    .9 o/o (processed),    .7 o/o (all)

None of the antirelay-logs was big enough to prevent it from being deleted. Most important is the amount of messages that originate from blacklisted sites.

There seems to be one procedure that generates errors – not fatal, appearantly – but it’s a bit of a mistery which one….

Hourly access
Since yesterday 17:00 UTC – being 19:00 CET+DST – there is a bunch of access on the server, repeating every 45-60 minutes. It shows up in both HyperSpy and T4: in all aspects: CPU, memory, DIO, BIO, pagefaults and number of processes. The WASD log cannot be accessed – or I have to cycle the logfile. But since the load has stopped early this morning, there is no9 direct need – it seems. Just that about 11:30 UTC, there has been a big spike in the number of processes (up to 80), paging (Duh) and BIO – but it did’t last long and the the load – in number of processes – dropp[ed to normal after 15 minutes or so. Again: no big deal…But worth looking at next week (when the logfiles becomes available)