Bugs

Symantec Definition date is stuck at December 31 2009

Most of our Symantec Endpoint Protection clients are alerting that the definition dates are old (we reduce the alert time to less than the default 30 days). These alerts are coming in through the desktop client and also through both of our monitoring systems. Apparently Symantec are aware of the issue (see “The date of the definitions in Symantec Endpoint Protection clients and Symantec Endpoint Protection Manager remain at Dec 31 2009”)  and their definitions cannot have a date in 2010. Therefore their work around is to push out new updates with a date of December 31st 2009 and they are just increasing the version number until engineering come up with a patch to fix the issue of not accepting dates in 2010.

I sure hope that their update plan works better than our most recent upgrade that meant we had to reinstall the client by hand at every desktop.  None of the upgrade processes would work.

Preparing Network Connections message at startup of SBS – solved.

Today of all days we’ve had two clients that have had their server reboot for a couple of valid reasons but after reboot the server just sat at “preparing network connections” screen and would not continue. We’re not sure right now what caused this issue but the solution was to reboot the server, press F8 and choose the Last Known Good to be able to get into the server.

Today has not been a good day for this to happen as some clients have been closed so they’ve not been around to let us in to look at the server on site but at the same time we don’t really want to wait until Monday to get access to the server, yet this is a holiday weekend.

For me, it’s been a long week . I’ve started work at 4am twice this week and was working at 1am until 2.30am last night so I doubt I’ll be staying up for NewYear – but I think I can make it until 7pm when I’ll be able to watch BigBen strike midnight.

Happy New Year everyone and I hope 2010 starts off better than 2009 finished!

Unable to rdp to Windows2008 SBS server from XP client after KB969084 installed

We had an issue when all of a sudden we were not able to remote desktop to a clients SBS 2008 server using the rdp client and the TSGateway functionality. Remote Web Workplace would work fine and so would Windows7 clients.
After proving this patch was the culprit by removing the patch and finding my saved rdp session would work, I went back and read the kbarticle 969084 on this patch. I hadn’t initially read this (in common with a lot of other people) and also because the patch was pushed down via wsus. It turns out that XP does not turn on CredSSP by default and this is needed to work with the new RDP client. I followed the instructions at kb951608 and after a reboot, going to the control box/About I got the message that Network Level Authentication was supported and I was then able to connect succesfully.
MSTSC showing Network Level Authentication Supported
To summarize you need todo the following.

  1. Click Start , click Run , type regedit , and then press ENTER.
  2. In the navigation pane, locate and then click the following registry subkey:
    HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Lsa
  3. In the details pane, right-click Security Packages , and then click Modify .
  4. In the Value data box, type tspkg . Leave any data that is specific to other SSPs, and then click OK .
  5. In the navigation pane, locate and then click the following registry subkey:
    HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\SecurityProviders
  6. In the details pane, right-click SecurityProviders, and then click Modify .
  7. In the Value data box, type credssp.dll . Leave any data that is specific to other SSPs, and then click OK .
  8. Exit Registry Editor.
  9. Restart the computer.

Imagemaps (navigation) broken in IE8 – fix

We rolled out IE8 to a customer earlier this week and promptly found their company website didn’t work in ie8 (despite some users having had IE8 for several months). An imagemap that they use for navigation did not show up in IE8 on internal computers. The weirdest thing is that all the computers at their office had the problem yet none of our computers or some other computers we tried could reproduce the problem.

After trying many technical solutions I passed it to our web developer who very quickly came up with a bug in ie8 and content produced by Publisher

“Publisher HTML output uses some very large numbers for object coordinates. This behavior has worked in the past. However, Internet Explorer 8 does not support such large coordinates. This is because some precision was moved from the most significant end to the least significant end of the coordinate variables to allow for sub-pixel layouts. Therefore, when large coordinate values in Publisher HTML output are run through Microsoft Dynamic HTML, the values are truncated. This behavior causes significant problems when Publisher HTML is rendered in Internet Explorer 8.”

Sure enough – saving the files within Publisher 2007 sp2 fixed the issue.

Alternative to windirstat?

We’ve used windirstat for a long time when investigating why server or desktop hard drives suddenly fill up. It’s a very quick way of finding out who has just synched their ipod to the server network share.  Earlier this week though we found that it was reporting a drive had 40gb used out of 115gb yet the OS was showing 2gb free. Obviously this is a rather large discrepency and was the second time we’d had windirstat not match the information that Windows was reporting.  After a bit of digging around, I checked the volume shadow copy settings, set it back to a more reasonable 10gb instead of 99.2% of the disk and our free space went back to about 70gb in windows – now matching what windirstat was reporting.

So it looks like windirstat does not take into account files such as the VSS cache location.

So we’re looking around for an alternative – I’d really like it to be a standalone application – if you know of anything then please add something in the comments and I’ll add a new post with my preferred solution.

Your Out of Office settings cannot be displayed, because the server is currently unavailable. Try again later – fixed

“Your Out of Office settings cannot be displayed, because the server is currently unavailable. Try again later” occurs when trying to access out of office onwith outlook2007. The strange thing is that the out of office functionality through the Outlook Web Access page works as expected.
There are several documented ways to fix this, mainly ensuring that the various autodiscover urls are correct. See Proexchange.be – Your out of office settings cannot be displayed for the best document on this.
Interestingly is that if you enable debugging in outlook and try to access the Out of Office you do see the settings being pulled across in the logfile.

However I was still having this issue. From Microsoft forums on Exchange Server Clients I found that various patches to the dot net framework (oh how I hate thee) being discussed and http://support.microsoft.com/kb/952883 was the first patch that was discussed. Sure enough, installing this patch fixed the problem and what is more I didn’t even have to reboot.

The annoying thing is that the first time I had this problem (on this server) was due to a typo in the autodiscover service, then the .net framework patches were applied and the problem re-occured.

Nothing happens when clicking on the start menu – fixed.

A while back we had a client that was migrating out of their existing domain and into a new SBS2008 installation. One of the things I learnt (too late) was to disable folder redirection before doing a migration otherwise clients will still point to the old server. Unfortunately I did not have access to the old server/domain but I had got a copy of the redirected folders and thankfully there was no real data on the server to be migrated from the redirected folders.

However to fix the redirected folders I had to use csccmd to remove references to the old server which was easy enough using “csccmd /unpin2:\\oldserver\share /recurse”. I then changed the registry entries in HKEY_CURRENT_USER \ Software \ Microsoft \ Windows \ CurrentVersion \ Explorer \ User Shell Folders to point to the new location.  This worked fine for every machine except one.  This particular machine had a very strange symptom in that when clicking on the Start/All Programs button the machine would pause for about 20 seconds and then nothing would happen (to the end user). Behind the scenes the Start Menu, Startup and Programs entries in the registry would get deleted. I tried to use ProcessMon from sysinternals to monitor the registry setting but I either got too much registry information to work out what was going on or nothing at all (depending on the filters I had applied).

Anyway, yesterday I stumbled across Ramesh’s site which mentioned running “regsvr32 /i shell32.dll”. I tried this, clicked on the All programs and nothing happened – again. I rebooted and the problem persisted. I then ran it again and was about to reboot the machine again when the user logged into the machine so I had to stop work (I was doing all this remotely using rdp). I logged into the machine this morning and checked the registry. Somehow the registry items were no longer blank but were repointing back to the original server. I reset them back to the new locations and now the All programs button works as designed. I think the trick was to run regsvr32 and then reboot before clicking on the All programs button. (Either that or reboot twice and then check the registry settings and correct them)

Firefox slow, especially gmail? – remove skype extension.

Ever since I updated the home computer to run the RTM version of Windows7 my gmail and greader experience has been very frustrating in Firefox. GReader just had no idea what data to display in the right hand side of the screen and gmail would just stick in the “please wait”, “loading” or “still working” status after sending an email (but would actually send the email).

Rebooting firefox into safe mode, or running ie8 worked fine so I knew it was not the computer but had to be an extension. Fortunately this was 100% reproducible so I was able to do some diagnostic tests.

I disabled all the likely culprits – flashblock, adblock,greasemonkey, better gmail, web developer, noscript but I still had the same problem. It wasn’t until I disabled the skype extension that the performance of google reader was back to normal and I could now send mail again.  Seeing as though I only really use Skype to call my parents I don’t need the extension so it is now disabled. For some reason the uninstall button is disabled.

I did find a thread on the Google Help forums with the same solution which was nice to know that I wasn’t the only one with the problem and I came up with the same solution.

MySQL installation error – The wizard was interrupted before MySQL Server 5.1 could be completely installed

I had to install a WordPress installation on IIS last week (I really would not recommend this is done – a linux install is so much easier) and part of that installation requires the install of MySQL for the database. The most current version is 5.1.35 but unfortunately there is a big bug with the install. It completely fails to install with the error message of “The wizard was interrupted before MySQL Server 5.1. could be completely installed.” With no useful explanation in the dialog box, event viewer or installation log files it is extremely hard to work out what is wrong. The solution for the time being is to download the 5.1.34 installation. The windows version is available at filehippo. It doesn’t help that the official MySQL web page does not link to previous versions of the software either. After logging a “me too” comment on what now seems to be the official bug report at MySQL Bugs – #45418, the number of people with the same problem has increased dramatically.
I had to register with the MySQL website to log my bug and they sent me an email a couple of days later mentioning the release of the new software (that has the problems) – rather annoying as it’s not a good idea to publicize software that doesn’t even install. I tweeted my comment that they should fix the issue before emailing out and I got a reply that they are aware of the problem and are working on it. No post has been added to the bug report though.

Dynamics SL crashing on startup after installation of .net patches.

Overnight several of the .net patches were pushed out to workstations and now when Dynamics SL  is run I get the following error. “Microsoft Dynamics SL has encountered a problem and needs to close. We are sorry for the inconvenience”

Looking in the event log or the details  I see a .net Runtime 2.0 error, Event 5000. Description EventType clr20r3, P1 msdynamicssl.exe, P2 7.0.0.0, P3 4889f859, P4 solomon.kernel, P5 7.0.0.0, P6 4649b3d9,P7 1b4, P8 0, P9 system.accessviolationexception, P10 NIL

This is then followed by id 5001, Description “Bucket 300554837, bucket table 5, EventType clr20r3, P1 msdynamicssl.exe, P2 7.0.0.0, P3 4889f859, P4 solomon.kernel, P5 7.0.0.0, P6 4649b3d9,P7 1b4, P8 0, P9 system.accessviolationexception, P10 NIL

The cause of this was actually the installation of a .NET Framework 3 patch.

The sorry saga of trying to fix this solution is documented at the dynamics forum but this is a known issue and the patch is available through Customersource or PartnerSource if you have access to this. Alternatively call Microsoft hotfix line on 1 888 456 5570 and request the Dynamics SL hotfix 961802. There is no charge for this hotfix.  Install on your workstation and you should be good to go.

However if you have tried to fix this issue by editing your .NET framework settings then I recommend totally removing .net from the computer, reinstalling v2 and the service pack.  There is a really good tool for removing .NET framework from Aaron Stebner which works a treat.