Wilkommen, Bienvenu, Welcome... Sziasztok!

Welcome to The Lotus Position, an intermittent collection of extempore navel gazings, ponderings, whinges, whines, pontifications and diatribes.

Everything is based on a Sample of One: these are my views, my experiences... caveat lector... read the Disclaimer

The Budapest Office - Castro Bisztro, Madach ter

The Budapest Office - Castro Bisztro, Madach ter
Ponder, Scribble, Ponder (Photo Erdotahi Aron)

Guest Nutter/Kindred Soul: Bill Bailey


Wednesday, 23 April 2008

A Bug! Fie on it!

I know it's putting one's head in the lion's mouth to adopt a new operating system as soon as it's released, but my old previous machine - a pathetic HP Pavilion [ab initio defective DVD, crashy, screen died at 1 year warranty + 1 day, etc. etc. etc.] - just had to be replaced and I hung on as long as I could...

In the end I was actually very pleased with Vista (Home Premium) - the Rock Xtreme CTX pro has plenty of horse-power and elephant-memory to support Vista and everything was absolutely fine. Some things omitted from Vista Home Premium were absolutely ridiculous omissions, I thought , but it ran nicely and very solidly (even allowing for the fact that it has nVidia Go 7950GTX graphics - and we all know about nVidia drivers now, don't we?).

There were a few hiccups along the way (VS2008 for instance) but the basic system was Rock solid...

And then along came SP1 - after which every day was a Blue Screen Day! Yes - every day a STOP ERROR "DRIVER_POWER_STATE_FAILURE" 0x0000009F error when I moved the laptop from home to the cafe where I write. SP1 actually installed OK, but it was after I had fixed the newly introduced nVidia drivers issue that the real problem became apparent.

I've now been in regular touch with Microsoft Technical Support for over a month now (though I haven't heard from them for a week), emailed other technical types who run blogs and so forth but the issue is still not resolved... though I hasten to assure you, dear reader, that my system no longer crashes.

Eh? Issue not resolved but not crashing? How come? Simple - turn off the offending piece of Microsoft software - the issue hasn't been resolved but it can be avoided. As of this moment the system has been running continuously for 958,054 seconds (that's 11 days, 2 hours, 7 minutes and 34 seconds). Not bad.

But what is this obscure issue that has so far apparently defeated Microsoft. Obscure - of course.

As noted here I had added a little bit of extra storage (1.3TB of RAID5 to be precise) and set up a persistent iSCSI connection (for technical reasons I shan't bore you with this was Necessary... couldn't do what I needed to do with a simple drive mapping, etc.) It was absolutely fine - until...

SP1. It seems something changed: if an iSCSI connection is set up as persistent, it seems that even an Administrator cannot log-off a session (if it's not persistent logging on and off works just fine). But! If - post SP1 - the laptop is put to sleep, disconnected from the network, moved, woken and reconnected to another network something goes very badly wrong... with the iSCSI port, the Virtual Disk complains and it all goes titsup. I didn't change how I worked, but one day Vista was happy and the next day it had developed epilepsy.

Took me a while to figure this out...I had to download a Microsoft debugging tool (windbg); I had to learn how to use it; I had to decipher the dumps, I had to learn all about IRP's (I/O Request Packet)s... (I know... me me me me me) - and then I had to communicate with various people and still largely do the job for myself (I hasten to add however that all of the people I contacted have been surprisingly willing to help - I'm surprised easily by helpful email support- although unfortunately unable to help... thus far).

It took me three weeks - losing at least half an hour a day to nail this - and I'll go so far as to say, dear Mr Microsoft, that I really wouldn't object to some tangible expression of sympathy (e.g. free upgrade to Ultimate or license to run the existing system under Virtual PC which would have been/may be handy for debugging purposes)

[Suggestion: MS should do this automatically anyway once an incident crosses a certain threshold of severity - no one wants to be without a primary work tool, so clean installs, etc. etc. are (they were for me) completely out of the question... but a lot could have been achieved in a virtual machine and it would go some way to compensating users without actually costing MS anything really. If you are tempted to mentioned e.g. System Restore, please don't.]

The bottom line is: if you are experience regular BSOD's with bccode 0x9F and you use iSCSI and you move the machine just before each crash - check to see if the iSCSI is a persistent connection... if it is, turn it off and see if the problem goes way. (There are similar issues recorded in the MS Knowledgebase for Windows Server 2003 but nothing for Vista)

The sad thing is that Microsoft has still been unable - or unwilling - to acknowledge this issue, and I think a month is perfectly adequate to get that far - a fix might take longer but such a reproducible BSOD should be gently squeezing someone's gonads... just cos you don't log off a connection the whole damn machine shouldn't crash!

And now, some Mozart...

Classic Stuff.

Wednesday, 2 April 2008

Long Thing Shortened

I just wanted to point out that after approximately two hours of concentrated effort and much contorted thinking this morning (in an endeavour to re-examine the underlying concept from a more illuminating angle) I have come to the conclusion that there is no single word in English for a "Long thing that is now much shorter than it was... but still quite long."

Nor could I think of any short and elegant way of referring to the long, but relatively short, remnant of something that was once much longer.

These are, I feel terrible omissions.

Likewise I have on occasion felt the need for - and been disappointed in my search for - a single word for:
  • The back of the hand, and
  • The hollow on the inside of the elbow
No doubt some things don't need to be referred to very often and so don't get their own word (inside elbow hollow for instance) but that can't be the whole story otherwise historians would be scouring the records of French history in search of a hitherto unrecorded plague of one-eyedness leading to the french word "borgne". Eh? I think that last bit was the wine talking.

Anyway, at least we have a word for Stuff.

Vista SP1 - BSOD, Bugs and Aggravations Part Deux

Does your system seem sluggish? Do you have one of those dashboard gadgets that says how much memory you are using and its seems to be saying "A Lot!"?

Or more specifically, have you checked the Processes tab of the Task Manager and seen row after row of taskeng.exe processes... tens? Hundreds? Thousands?

It pays to check occasionally. I only had a few hundred taskeng.exe processes, but that's more than enough to say WTF!

Quick Summary...

Taskeng.exe is the process that shows up for tasks created by the Task Scheduler. In Task Scheduler you can see which process belongs to which task by showing the Preview Pane and selecting a task event on the History tab and looking at the info on the General tab just below, where you will see something like

Task Scheduler launch task "\Start AVG Control Centre" , instance "C:\Program Files\Grisoft\AVG7\avgcc.exe" with process ID 2424.

But, if you have the particular problem of taskeng overload you won't find any corresponding tasks or events unless you select "Show Hidden Tasks" from the View menu... and then voila!

What you should probably be looking for is a task in Task Scheduler Local Library called "User_Feed_Synchonization... This task takes care of RSS feed collection, so if you don't use RSS feeds at all you will probably never have this problem, but if you do, look at its History... lots of failures? Yep, "UFS" is spawning another instance of itself every 5 minutes by default, and if they don't also terminate automatically that's an accumulation of 288 a day, each consuming a bit less than 1MB. Leave your machine on for a few days without restarting and >95% of the processes will be doing nothing helpful at all.

To resolve the problem - which is that the task has been corrupted - delete all existing "User_Feed_Synchronization..." tasks and then open a Cmd window.

Type "MSFEEDSSYNC DISABLE" (belt and braces approach) and then "MSFEEDSSYNC ENABLE", which will create a new feed collection task. The default task runs Once at 01:45 and then every 5 minutes until the next midnight... and then every five minutes thereafter so feeds won't be collected until either 01:45 or you kick start the task...

But at least you will own (a bit more of) your PC again

More info on checking the RSS User_Feed_Synchronization task, registry entries etc. from a Cmd window can be found on the Microsoft RSS blog, and my thanks to the guys, especially Andrew208, in the Technet forum who initially sorted this out in the post Re: Solved: 1055 spawned taskeng.exe threads... WTF? It seems this is an SP1 RC1 bug that didn't get stomped properly.

One of the few occasions on which one finds a problem and almost immediately a full answer... Nice Stuff.

Tuesday, 1 April 2008

Vista SP1 - BSOD, Bugs and Aggravations

I'm fed up with this particularly useless piece of "assistance" from Microsoft. Nice idea the "Problem and Solution Centre", especially since this Vista HP laptop is BSOD'ing every day now, so I thought I'd check to see what it had to say about the current state of my somewhat poorly system...

Aha! Let's follow that link and follow those obviously helpful steps...

What driver? Which device? How about a link to the Knowledgebase?

I have lost count of the times I have clicked the "No" brackets- this-was-not-helpful-at-all-quite-the-reverse-thank-you-very-much- if-you-want-to-add-my-blood-pressure-to-the-list-of-problems-now- close-brackets button and conveyed my displeasure (in decreasingly moderate terms of course) in response to the helpful question "Was this information helpful?", so maybe, just maybe, a little humiliation is in order.

Yes, it's in order - but whether it will help or not is another issue.

(BTW, the laptop manufacturer reliably informs me that the BIOS does not need to be updated and rather wished that MS would stop suggesting this as it creates an inordinate amount of completely unnecessary work for them...)

And, just to round things off, in case you are ever worried about whether your Drivers are properly signed and aren't entirely certain, please be aware that neither is Vista, so don't expect much help there either...

So, as you can see snapman.sys (of which there is only one on the system) is both signed and not signed... maybe Acronis signed for the fact that they weren't signing it? Who knows, but I see the imprimatur of MS behind it all.

By the way... the BSOD's are all 0x9F "DRIVER_POWER_STATE_FAILURE" errors and I suspect MS again with regard to the USB drivers: instead of just dying some 10-20 minutes after plugging in my Lexar JumpDrive Lightning II drive when I plugged it in today Vista refused to acknowledge the existence of the device at all then crashed when I took the non-existent device out again...

Of course everything was just dandy before SP1... but, 'tis true - plus ca change...

As as I saw in a forum sig recently "The mouse has been moved - please restart your system to apply the new settings"...

Annoying Stuff