Copying certain files keeping directory structure with Robocopy

Leave a reply

I have a bad habit when it comes to storing files. When I reinstall an operating system, or upgrade a hard drive, I keep a copy of the existing hard drive (usually in a very descriptive directory named something like c).

Sometimes I might remove some of the unnecessary files, if there is time…

Over time, these ‘backups’ become nested, so I end up with a monstrosity like c:\users\aaron\desktop\c\users\aaron\desktop\c_backup\users\aaron\desktop\phone_backup\photos….

Awesome!

Windows 10 comes with a nifty and powerful copying tool called Robocopy. Say you wanted to pull out all the .jpg files within a disk or directory, but keep the directory structure (so that if files have the same name they don’t overwrite):

robocopy  [destination] [file filter] /S

(/S does not create empty directories. If you want to keep them use /E instead)

e.g. If you wanted to copy all .jpg files from g: to m:\old_photos, the command would be:

robocopy g:\ m:\old_photos\ *jpg /S

Applying Andon to software development

Leave a reply

Brain dump, may make this more coherent later

Andon is a manufacturing management process for quickly dealing with critical issues. When a quality or process issue is found such as defective or missing parts, incomplete or incorrect installation, or issues that block a process, a worker can immediately escalate the issue to 1 or more senior staff who must immediately respond. If a fault is not rectified within a certain time, or an automated system is tripped, the manufacturing line is halted until the issue has been dealt with.

The key benefit of Andon is that it allows for critical issues to be escalated and examined within minutes, and brings in the ‘experts’ who have the knowledge to establish how to best respond to the issue. Electrical power system manufacturer Ensto has dedicated mobile phones for the response team, and all members must respond to an escalation within 5 minutes.

For software, relying on an issue to be escalated through usual channels can be too slow for dealing with truly critical problems.

Common issues in software development

  • A defect is found by a tester or reported to 1st level support. They suspect that the defect is serious, but don’t know who is responsible. The defect is added to an escalation queue. The issue gets passed around until someone understands the impact.
  • During a final meeting before a build it is announced that a build must be postponed due to a critical issue that only a couple of managers were aware of.
  • Work done by one team has broken functionality in another area. The team is ordered to fix the problem they caused.
  • A build is delayed because a developer working on a critical issue is sick.

 Using Andon

It’s unlikely that you would want to completely drop Andon into your development process (although flashing alert lights and completely stopping production would probably get critical issues resolved quickly).

Some of the concepts though can be used to improve how you react and deal with critical issues.

Build a response team

Unless it is a small product you really need a small team of experts from a number of areas. The team is notified of critical issues and expected to investigate them quickly. Between them they should be able to establish if the issue is truly critical, and ideally get it fixed right now.

Notifying all response team members keeps everyone up to date on any current critical issues.

Multiple team members bring multiple views on an issue. Too often a critical issue is discounted because an individual doesn’t fully understand the impact of the issue, while having a number of different viewpoints can prevent very serious problems being dumped back in the standard queues to be investigated.

Having a team means that a single staff member away does not prevent a critical issue being fixed.

Drop everything

Critical means that this issue needs to be resolved now. Realistically truly critical issues will be infrequent, and it should be expected that the response team literally stops everything they are doing to look at the raised issue and get it fixed.

Define critical

For manufacturing this can be because all the outputs are going to be defective. For software, this depends on the company and can be anything from system completely dead, important functionality not working or top-level client cannot perform a required process. Care needs to be taken to not make the scope too broad as each critical event is disruptive to everyone on the response team. Any changes to scope need to be clearly and widely communicated to ensure that all staff are aware of what they need to immediately escalate.

Stop wasting time

Most staff are not experts on the intricacies of your software. If someone finds an issue that they suspect is critical, it makes sense for it to be immediately passed to someone who can confirm if it is critical AND who can get the issue fixed. Worst case is that the issue is not critical and the escalation team spends a short period confirming this. Excessive false alerts from new staff can be mitigated by having them check with a colleague when possible.

Stop the blame game

When Team A does work that catastrophically breaks Team B’s area, this now everyone’s problem to fix. A small team of experts is likely to be quicker in investigating and repairing a critical issue than the issue being first sent to Team B to look into, who then blames Team A and expects A to fix.

Review the causes

While not Andon specifically, every critical issue should be examined to determine how it occurred and if/how it can be avoided. You don’t want the same problem to occur again. Whether you make your system more resilient, implement auto tests or restrict access to key codebases, it’s good to be able to demonstrate that you have done everything you can to improve the reliability of your system.

Windows 10 does not respond to ping

Leave a reply

By default, a new Windows 10 installation will have the firewall enabled and set to not respond to ICMP pings, even from hosts on a private network. This can resolved enabling the correct whitelist entries in the firewall.

  1. Click the Windows menu icon, then start typing firewall to start listing programs. Select Windows Firewall with Advanced Security.

windows 10 ping image 1

  1. Select Inbound Rules (as we want the computer to respond to incoming pings).
  2. Find the rules File and Printer Sharing (Echo Request). There will be rules for IPv4 and IPv6, and possibly for different network Profiles. Right-click the rule/s you wish to activate and select Enable Rule. Most internal and home networks currently run IPv4.

windows 10 ping image 2

As soon as the rule is enabled the computer should start responding to pings.

Setting up headless Ubuntu server in VMware Virtual Machine

Leave a reply

In order to have a headless (no screen) Ubuntu vm running in the background in Windows, first you must install and setup Ubuntu, then launch the virtual machine using vmrun with the nogui option. Below sets up Ubuntu for both ssh and VNC remote desktop access.

  1. Install Ubuntu.
  2. Setup SSH server (Secure shell client and server metapackage).
  3. Set the resolution.
  4. Enable VNC remote desktop: Applications > Desktop Sharing:
    1. Tick Allow other users to view your desktop.
    2. Tick Allow other user to control your desktop.
    3. Untick You must confirm each access to this machine (don’t leave this ticked, otherwise you need to already be viewing the desktop in order to access remotely).
    4. Tick Require the user to enter this password, then enter a password.
    5. Click Close.
  5. Confirm that you can log in to Ubuntu via SSH and/or VNC.
  6. Shutdown virtual machine.
  7. Open the virtual machine from the command prompt with vmrun and the option nogui.

e.g. c:\Program Files (x86)\VMware\VMware Workstation\vmrun start m:\vm\Ubuntu\Ubuntu.vmx nogui

Enable the CTRL/Control key in VMware Fusion

Leave a reply

The default settings in VMware Fusion set the Control key as a mouse modifier, where holding Control and clicking is treated as a right-click. This makes it impossible to CTRL select files in virtual machines.

This can be disabled by:

  1. In VMware Fusion access the VMware Fusion menu, then select Preferences.
  2. Access Keyboard & Mouse.
  3. Select Mouse Shortcuts.
  4. Untick Secondary Button. Alternatively, you can double-click the option and change the keyboard key assignment, but I prefer just to turn it off.
  5. Close the preferences.

The World’s Best Banana Cake

Leave a reply

Ingredients

  • 125g butter (if unsalted add 1/2 teaspoon salt to flour)
  • 330g sugar
  • 280g mashed banana (avocado masher works well, or a fork and a cup/bowl)
  • 1/2 teaspoon vanilla extract
  • 2 eggs
  • 100ml buttermilk
  • 225g self-raising flour
  • 1/2 teaspoon bicarb soda

Method

  1. Preheat oven to 180°C.
  2. Cream butter and sugar.
  3. Add banana, vanilla and eggs. Mix. Suggest not using a whisk (I had to try out my fancy new whisk from Amazon last time I made this, and the banana tends to get stuck).
  4. Mix in the buttermilk.
  5. Sift the flour and bicarb soda together (and salt if using unsalted butter). Add to wet ingredients and mix.
  6. Dump mix in a loaf pan, cook for around 50 minutes.

Delete hiberfil.sys on Windows 7

Leave a reply

If you don’t use hibernation, hiberfil.sys is a waste of harddrive space. It stores the contains of RAM when the system hibernates, so is the same size as your system RAM. In my case it was taking up over 8GB of space.

  1. Open the Start menu, type in Command, right-click on Command Prompt and select Run as Administrator.
  2. If prompted to confirm click Yes.
  3. Type in powercfg.exe -h off, then press ENTER.

Done. Hibernation is disabled and the file is automatically removed.

Dealing with Windows 7 Time Sync issues

Leave a reply

For reasons (I suspect the CMOS battery, but have no replacements), the clock on my Windows 7 PC ceased working correctly. On boot, the time is very close to the time the computer was last shut down, which is often greater than 12hrs behind. It does not automatically pick up the correct time.

I initially thought that this was because there are limits to the maximum time difference when syncing, so that if the clock is too far out the system time is not updated. I tried overriding these limits in the registry, however as the default sync only runs once a week this isn’t a complete fix. My temporary solution (until a new battery is obtained) is to also set up a resync Schedule Task for each boot.

Step 1. Overriding the registry sync limits

  1. Open Regedit.
  2. Navigate to HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\W32Time\Config.
  3. Change MaxNegPhaseCorrection and MaxPosPhaseCorrection from d2f0 to ffffffff.
  4. Reboot.

Step 2. Setting up the Scheduled Task

  1. Open Task Scheduler.
  2. In the left panel find and highlight Task Scheduler (local) > Task Scheduler Library > Microsoft > Windows > Time Synchronization.
  3. In the Actions panel (right side) click Create Task….
  4. Enter a Name, e.g. Force sync time.
  5. Click Change User or Group….
  6. Click Advanced….
  7. Click Find Now.
  8. In the Search results list select LOCAL SERVICE., then click Ok and Ok.
  9. Tick Run with highest privileges.
  10. In Configure for select your operating system (why this doesn’t default to Windows 7 I don’t know).
  11. Activate the Triggers tab.
  12. Click New….
  13. Enter how often the Task should run. I set it to: At Startup with 1 minute delay. (I tried without a delay, and it seemed to try to update before the network was up and fail).
  14. Activate the Actions tab.
  15. Click New….
  16. Leave Action as Start a program.
  17. In Program/script enter %windir%\system32\sc.exe. In Add arguments enter start w32time task_started. This ensures the Windows time sync service is running.
  18. Click Ok.
  19. Click New… again, but this time enter a new Start a program with Program/script of %windir%\system32\w32tm.exe with and argument /resync. This synchronises the time.
  20. Activate the Conditions tab.
  21. Tick Start only if the following network connection is available, and leave it set to Any connection (as you need a network connection to synchronise the time). Optionally you can also untick the 1st Power option, as the Task won’t use much  power when using battery.
  22. Activate the Settings tab.
  23. Tick Run task as soon as possible after a scheduled start is missed.
  24. Click Ok.

You can now test the Task by clicking Run. If you clock time was incorrect, it should now be correct.

If the clock issue is that it loses time while running, this could indicate more than a battery problem. A quick fix would be to set up a Scheduled Task as above, but set it to run constantly such as every 15 min.

Undoing the changes

You can easily undo the changes above, by setting the Registry values back to the original values and deleting the Scheduled Task.

RoboHelp 7 webhelp does not work in Chrome

Leave a reply

The RoboHelp 7 webhelp output is hard coded to check what browser you are using (don’t get me started on this, I hate user agent checks being used to disable functionality). As RH7 predates Chrome, webhelp doesn’t work. Specifically, the Table of Contents (ToC), Index, Glossary and Search does not load.

You can resolve this by editing the whver.js file in the output. The one downside is that this file is regenerated on each build. If you use software to sync the output files to a server you should be able to exclude whver.js from being updated when uploaded new builds.

3 changes to the whver.js file are needed:

// WebHelp 5.10.006
var gbNav=false;
var gbNav6=false;
var gbNav61=false;
var gbNav7=false;
var gbNav4=false;
var gbIE4=false;
var gbIE=false;
var gbIE5=false;
var gbIE55=false;
var gbOpera6=false;
var gbOpera7=false;
var gbKonqueror3=false;
var gbSafari3=false;
var gbChrome = false ;

var gAgent=navigator.userAgent.toLowerCase();
var gbMac=(gAgent.indexOf("mac")!=-1);
var gbSunOS=(gAgent.indexOf("sunos")!=-1);
var gbUnixOS=(gAgent.indexOf("linux")!=-1) || (gAgent.indexOf("unix")!=-1);
var gbOpera=(gAgent.indexOf("opera")!=-1);
var gbKonqueror=(gAgent.indexOf("konqueror")!= -1);
var gbSafari=(gAgent.indexOf("safari")!= -1);
var gbWindows=((gAgent.indexOf('win')!= -1)||(gAgent.indexOf('16bit')!= -1));
var gbMozilla=((gAgent.indexOf('gecko')!=-1) && (gAgent.indexOf('netscape')==-1));
var gbChrome=(gAgent.indexOf("chrome")!=-1);

var gVersion=navigator.appVersion.toLowerCase();

var gnVerMajor=parseInt(gVersion);
var gnVerMinor=parseFloat(gVersion);

if(!gbOpera&&!gbKonqueror&&!gbSafari) // opera can mimic IE or Netscape by settings.
{
gbIE=(navigator.appName.indexOf("Microsoft")!=-1);
gbNav=(gAgent.indexOf('mozilla')!=-1) && ((gAgent.indexOf('spoofer')==-1) && (gAgent.indexOf('compatible')==-1));
if(gnVerMajor>=4)
{
if(navigator.appName=="Netscape")
{
gbNav4=true;
if(gnVerMajor>=5)
gbNav6=true;
}
gbIE4=(navigator.appName.indexOf("Microsoft")!=-1);
}
if(gbNav6)
{
var nPos=gAgent.indexOf("gecko");
if(nPos!=-1)
{
var nPos2=gAgent.indexOf("/", nPos);
if(nPos2!=-1)
{
var nVersion=parseFloat(gAgent.substring(nPos2+1));
if(nVersion>=20010726)
{
gbNav61=true;
if (nVersion>=20020823)
gbNav7=true;
}
}
}
}else if(gbIE4)
{
var nPos=gAgent.indexOf("msie");
if(nPos!=-1)
{
var nVersion=parseFloat(gAgent.substring(nPos+5));
if(nVersion>=5)
{
gbIE5=true;
if(nVersion>=5.5)
gbIE55=true;
}
}
}
}
else if (gbOpera)
{
var nPos = gAgent.indexOf("opera");
if(nPos!=-1)
{
var nVersion=parseFloat(gAgent.substring(nPos+6));
if(nVersion>=6)
{
gbOpera6=true;
if(nVersion>=7)
gbOpera7=true;
}
}
}
else if (gbKonqueror)
{
var nPos = gAgent.indexOf("konqueror");
if(nPos!=-1)
{
var nVersion = parseFloat(gAgent.substring(nPos+10));
if (nVersion >= 3)
{
gbKonqueror3=true;
}
}
}
if(gbSafari)
{
var nPos = gAgent.indexOf("version/");
if(nPos!=-1)
{
var nVersion = parseFloat(gAgent.substring(nPos+8,nPos+9));
if (nVersion >= 3)
{
gbSafari3=true;
}
}
}
if (gbChrome)
{
gbSafari = true ;
gbSafari3=true;
}

var gbWhVer=true;