Related
I just finished (mostly) a major application that I've been working on for a little over a year (off and on). It is around 86k lines of code, 50k of those is from Visual Studio's auto-generated dataset. It's largely a GUI to interacting with the database, generating reports, etc. It deals with money and manages contracts so it is important for it to be as bug free as possible.
I've walked through the code, and ran the program myself. I, for the most part, cannot find more bugs. I am however, sure there are, I've just been working on the system so long I can't see them anymore. I know there are some, because of some intermittent issues I run across, but can never pinpoint.
How should I go about software testing in order to discover the remaining bugs?
I know this is a little late, but have you heard of Test-Driven Development?
There are lots of tests you could build to discover the "remaining" bugs:
1) Unit tests
2) Integration tests
3) Behavior/Business/Acceptance tests
You could always attend a Developer Testing Bootcamp to get more ideas.
You can involve some of your end users and do a beta test that way. The less experience they have with the application until now, the more likely they are to try things you didn't think of.
Since you didn't use TDD to write it. Your best bet now is to add as many automated tests as possible to cover common scenarios. That way, when you do find bugs, and there are ALWAYS bugs in programs, you can hopefully minimize the risk to the rest of the system when you fix them.
As one who is relatively new to the whole support and bug fixing environment and a young programmer I had never come across a bug that only occurs in the Websphere environment but not on the localhost test enviroment, until today. When I first got this bug report I was confused as to why I couldn't reproduce it on the localhost test environment. I decided to try on the Websphere test environment to see what would happen and I successfully reproduced the bug. The problem is I can't make changes and build to the Websphere test enviroment. I can only make changes to my local environment. Given this handicap what methodologies exist for resolving these kinds of bugs. Or are there even any methodologies at all? Any advice or help on how to approach issues like this?
Campaign for full access to a test environment. Being able to tweak things, redeploy and retry makes a huge difference. It's entirely reasonable to explain how not having access severely restricts your ability to do your job.
Make sure you've got sufficient logging, and make it configurable. Make sure you keep the logs for long enough to track down a problem reported by a customer even if it happened a few days ago.
When you finally diagnose a problem and why it only happens in a particular environment, document it and try to persuade your local system to behave the same way - that should make it easier to diagnose another symptom of the same problem next time.
In short, the methodology is to isolate and understand the differences between environments and which one or ones might be causing the issue.
Check your local build. Make sure it the same version (code and database) as Test and Prod. If it is, what are the environment differences that could effect the issue you are seeing? (Multi-core, load balancing, operating system version, 3rd library version). Don't run locally in the debugger, make sure your running a release build (if that's what is on Test and Prod) and maybe even do a local deployment rather than building from source.
Check to see if it is particular data that might be causing the problem. If you can, pull a copy of the database back from Test onto Local and see if that enables you to repro the problem.
Check with other developers. See if they can repro. the issue in their environment. Check with your QA guys, get their thoughts on what might be causing such an issue (often times they will have seen "similar" issues and might give you a clue).
At that point, if nothing helps, I generally go into a deep state of zen to try and understand what I am missing. But, there must be a difference, you just have to find it.
The Scientific Method always applies-- check your assumptions first. If the systems are different, the problem might reside in some sort of implicit default being different, or a different implementation of some function.
In all debugging processes, localization is the key. You gotta isolate the area of the problem first. If your OS, patches, libraries, and the main software itself are all identical, then it's probably the system settings (limits for sockets, file descriptors, etc). If you know you have enough inodes, space and memory left, then it's not a resource issue. If the computer is barely responsive to your interactive prodding, your load is too high, or you have some runaway processes. Remember what every process needs to run, and make sure they got what they need.
It can be also code just can't deal with the load of the production system. Locking mechanisms are a very popular cause of problem in production vs dev/test systems, simply because you can't generate enough of test cases that you get for free in production.
Logging can be easily overlooked, but I also like to add a lot of debug values into the code, to make debugging easier. I cannot even count how many times some environment variable, path, or broken symlink have ruined my day, just to realize that it would be a trivial fix if I looked at the values of variables while running, not just looking at the static code.
If all else fails, ltrace and strace are the best way to really look at what's going on under the hood. They're not easy to read, but once you get used to how spot and interpret return values of some syscalls, you gain a very powerful debugging tool.
Obviously the non-reproducible bugs are the hardest to fix due to the nature of their cause (i.e. race conditions), so we as programmers must do our best to gather data (i.e. logs, screenshots, etc.) and verify the bug documentation is accurate in an attempt to understand what happened. Can anyone recommend any software tools, or methods, that can record and reconstitute the actual executed sequence of machine instructions while allowing the user to step through them and inspect the code?
If it helps, the project I'm building is a windows application written in C++ and uses VS2005.
Thanks in advance for all your help!
'Time-machine' / Replay debugging is very helpful for debugging the type of issues you describe.
eg Green Hills time machine debugger
I have not used this myself but it sounds like it might be useful for the type of project you are building: VMWare replay debugging
Very occasionally, despite all testing efforts, I get hit with a bug report from a customer that I simply can't reproduce in the office.
(Apologies to Jeff for the 'borrowing' of the badge)
I have a few "tools" that I can use to try and locate and fix these, but it always feels a bit like I'm knife-and-forking it:-
Asking for more and more context from the customer: (systeminfo)
Log files from our application
Ad-hoc tests with the customer to attempt to change the behaviour
Providing customer with a new build with additional diagnostics
Thinking about the problem in the bath...
Site visit (assuming customer is somewhere warm and sunny)
Are there set procedures, or other techniques than anyone uses to resolve problems like this?
One of the attributes of good debuggers, I think is that they always have a lot of weapons in their toolkit. They never seem to get "stuck" for too long and there is always something else for them to try. Some of the things I've been known to do:
ask for memory dumps
install a remote debugger on a client machine
add tracing code to builds
add logging code for debugging purposes
add performance counters
add configuration parameters to various bits of suspicious code so I can turn on and off features
rewrite and refactor suspicious code
try to replicate the issue locally on a different OS or machine
use debugging tools such as application verifier
use 3rd party load generation tools
write simulation tools in-house for load generation when the above failed
use tools like Glowcode to analyse memory leaks and performance issues
reinstall the client machine from scratch
get registry dumps and apply them locally
use registry and file watcher tools
Eventually, I find the bug just gives up out of some kind of awe at my persistence. Or the client realises that it's probably a machine or client side install or configuration issue.
Extensive logging usually helps.
The easiest way is always to see the customer in action (assuming that its readily reproducible by the customer). Oftentimes, problems arise due to issues with the customer's computer environment, conflicts with other programs, etc - these are details which you will not be able to catch on your dev rig. So a site visit might be useful; but if that's not convenient, tools like RealVNC might help as well in letting you see the customer 'do their thing'.
(watching the customer in action also allows you to catch them out in any WTF moments that they might have)
Now, if the problem is intermittent, then things get somewhat more complicated. The best way to get around this problem would be to log useful information in places where you guess problems could occur and perhaps use a tool like Splunk to index the log files during analysis. A diagnostic build (i.e. with extra logging) might be useful in this case.
I'm just in the middle of implementing an automated error reporting system that sends back to me information (currently via email although you could use a webservice) from any exception encountered by the app.
That way I get (nearly) all the information that I would do if I was sitting in front of VS2008 and it really helps me to work out what the problem is.
The customers are also usually (sorta) impressed that I know about their problem as soon as they encounter it!
Also, if you use the Application.ThreadException error handler you can send back info on unexpected exceptions too!
We use all the methods you mention progressively starting with the easiest and proceeding to the harder.
However you forget that sometimes hardware is at fault. For example, memory could be malfunctioning and some computation-intensive code will behave strangely throwing exceptions with weird diagnostics. Of cource, it works on your machine, since you don't have faulty hardware.
Experience is needed to identify such errors and insist that customer tries to install the program on another machine or does hardware check. One thing that helps greatly is good error handling - when your code throws an exception it should provide details, not just indicate that something is bad. With good error indication it's easier to identify such suspicious issues related to faulty hardware.
I think one of the most important things is the ability to ask sensible questions around what the customer has reported... More often than not they're not mentioning something that they don't see as relevant, but is actually key.
Telepathy would also be useful...
We've had good success using EurekaLog with it posting directly to FogBugz. This gets us a bug report containing a call stack, along with related system info (other processes running, memory, network details etc) and a screen shot. Occasionally customers enter further info too, which is helpful. It's certainly, in most cases, made it much easier and quicker to fix bugs.
One technique I've found useful is building an application with an integrated "diagnostic" mode (enabled by a command line switch when you launch the app). That certainly avoids the need to create custom builds with additional logging.
Otherwise, it sounds like what you're doing is as good an approach as any.
Copilot (assuming customer is somewhere cold and rainy :)
The usual procedure for this is to expect something like this will happen and add a ton of logging information. Of course you don't enable it from the beginning, but only when this happens.
Usually customers don't like to have to install a new version or some diagnostic tools. It is not their job to do your debugging. And visiting a client for cases like these is rarely an option. You must involve the client as little as possible. Changing a switch and sending you a log file is OK - anything more than this is too much.
I like the alternative of thinking the problem at the bath. I will start from trying to find out the differences between my machine and the client's configuration.
As a software engineer doing webstuff (booking/shop/member systems etc) the most important thing for us is to get as much information from the customer as possible.
Going from
it's broke!
to
it's broke! & here are screenshots of
every option I picked whilst
generating this particular report
reduces the amount of time it takes us to reproduce and fix an issue no end.
It may be obvious, but it takes a fair amount of chasing to get this kind of information from our customers sometimes! But it's worth it just for those moments you find they're not actually doing what they say they are.
I had these problems also. My solution was to add lots of logging and give the customer a debug build with all the possible debug information. Then make sure dr Watson (it was on Windows NT) created a memory dump with enough information.
After loading the memory dump in the debugger I could find out where and why it crashed.
EDIT: Oh, this obviously only works if the application terminates violently...
I think following the trail of the actions user took can lead us to the reasons of failure or selective failures. But most of the times users are at loss to precisely describe the interactions with the applications, the automatic screenshot taking (if it is desktop app. for .net app you can check Jeff's UnhandledExceptionHandler). Logging all the important action which change state of the objects can also help us in understanding it.
I don't have this problem very often, but if I did, I would use a screen sharing or recorded application to watch the user in action without having to go there (unless, as you said, it's warm and sunny and the company pays the trip).
I have recently been investigating such an issue myself. Over the course of my carrier I have learnt that, while computer systems may be complex, they are predictable so have faith that you can find the problem. My approach to these kinds of issues two fold:
1) Gather as much detailed information as possible from the customer about their failure and analyse it meticulously for patterns. Gather multiple sets of data for multiple failure occurrences to build up a clearer picture.
2) Try and reproduce the failure in house. Continue to make your system more and more similar to the customers system until you can reproduce it, the system is identical or it becomes impractical to make it more similar.
While doing this consider:
1)What differences exist between this system and other working systems.
2)What has recently changed in your product or the customers configuration that has caused the problem to start occurring.
Regards
Depending on the issue you could get WinDbg dumps, they normally give a pretty good idea of what is going on. We have diagnosed quite a few problems that weren't crashed from minidumps.
For .Net apps we also was Trace.Writeline then we can get the user to fire up DbgView and send us the output.
Its very complicated issue . I was thinking writing some procedure for this . I just made some procedure for this non-reproducible bug . it might be helpful
When the Bug accorded .. There are several factors it might to occur.
I am Sure all bugs are reproducible . I always keep eye for these kind of issues..
Get the System Information
what other process the customer did before that.
Time period it occurs . its rare or frequent
its next action happened after the issue ( its always same or different )
Find the factors for this bug ( as developer )
Find the exact position where this issue happened .
Find ALL THE SYSTEM Factors on that time
check all memory leaks or user error issue or wrong condition in code
List out all facotrs to may cause this issue.
How the each factors are affected this and wat are the data is holding those factors
Check memeory issues happened
check the customer have the current update code like yours
check all log from atleast 1 month and find any upnormal operation happened . keep on note
Just a short anecdote (hence 'community wiki'): Last week I thought it was a clever idea in a Django app to import the module pprint for pretty printing Python data only if DEBUG was True:
if settings.DEBUG:
from pprint import pprint
Then I used here and there the pprint command as debugging statement:
pprint(somevar) # show somevar on the console
After finishing the work, I tested the app with setting DEBUG=False. You can guess what happened: The site broke with HTTP500 errors all over the place, and I did not know why, because there is no traceback if DEBUG is False. I was puzzled that the errors disappeared magically, if I switched back to debug mode.
It took me 1-2 hours of putting print statements all over the code to find that the code crashes at exactly the above pprint() line. Then it took me another half an hour to convince myself to stop banging my head on the table.
Now comes the moral of the story:
Not every thing that looks like a clever idea in the first view is such savvy in the end.
An important point to look at for debugging these errors are all configuration options and platform switches your code by itself makes. This can be quite a lot more than just some user preferences. Document good, if you make an assumption about the user's platform (e.g., if you test for Win/Mac/Linux only, will your code crash on BSD or Solaris?)
Cheers,
However tough a non-reproducible problem is - we can still have a structured and strategic approach to solve them - and I can say through experience that it requires out of box thinking in 50% of the cases. Generally speaking, one can categorize the problems into different types which helps to identify what tool to be used. For example if you have a non-reproducible application crash issue or a memory issue you can use profilers and nail down the issue caused in the particular functionality.
Also, one of the most important approach is inforamation rich logging. I also use a lot of enums to describe the state of the process depending on the scenario in question. for exampe, I used like Initiated, Triggered, Running, Waiting Repaired etc to describe the schedules states and saved them to DB at different stages.
Not mentioned yet, but "directed code review" is one good solution, especially if you didn't do a proper review (at least 1 hour per 100 lines of code) before release.
I have also seen impressive demos of AppSight Suite, which is basically an advanced environment monitoring and logging tool. It allows the customer to record what happens on his machine in an extensive but fairly compact log file which you can then replay.
As many have mentioned, extensive logging, and asking the client for the log files when something goes wrong. In addition, as I worked more with web apps, I'll also provide detailed, but succinct deployment documentation (e.g., deployment steps, environmental resources that need to be set up etc).
Here are common problems I've seen that lead to the types of problem you are describing:
Environment not set up properly (e.g., missing environment variables, data sources etc).
Application not fully deployed (e.g., database schema not deployed).
Difference in operating system configuration (default character encoding being the most common culprit for me).
Most of the time, these issues can be identified through the log content.
You can use tools like Microsoft SharedView or TeamViewer to connect to remote PC and inspect problem directly on site. Of course, you'll need cooperation with customer.
When working in a big project that requires debugging (like every project) you realize how much people love "printf" before the IDE's built-in debugger. By this I mean
Sometimes you need to render variable values to screen (specially for interactive debugging).
Sometimes to log them in a file
Sometimes you have to change the visibility (make them public) just to another class to access it (a logger or a renderer for example).
Some other times you need to save the previous value in a member just to contrast it with the new during debugging
...
When a project gets huge with a lot of people working on it, all this debugging-specific code can get messy and hard to differentiate from normal code. This can be crazy for those who have to update/change someone else's code or to prepare it for a release.
How do you solve this?
It is always good to have naming standards and I guess that debug-coding standards should be quite useful (like marking every debug-variable with a _DBG sufix). But I also guess naming is just not enough. Maybe centralizing it into a friendly tracker class, or creating a robust base of macros in order to erase it all for the release. I don't know.
What design techniques, patterns and standards would you embrace if you are asked to write a debug-coding document for all others in the project to follow?
I am not talking about tools, libraries or IDE-specific commands, but for OO design decisions.
Thanks.
Don't commit debugging code, just debuggin tools.
Loggin OTOH has a natural place in execption handling routines and such. Also a few well placed logging statments in a few commonly used APIs can be good for debugging.
Like one log statment to log all SQL executed from the system.
My vote would be with what you described as a friendly tracker class. This class would keep all of that centralized, and potentially even allow you to change debug/logging strategies dynamically.
I would avoid things like Macros simply because that's a compiler trick, and not true OO. By abstracting the concept of debug/logging, you have the opportunity to do lots of things with it including making it a no-op if needed.
Logging or debugging? I believe that well-designed and properly unit-tested application should not need to be permanently instrumented for debugging. Logging on the other hand can be very useful, both in finding bugs and auditing program actions. Rather than cover a lot of information that you can get elsewhere, I would point you at logging.apache.org for either concrete implementations that you can use or a template for a reasonable design of a logging infrastructure.
I think it's particularly important to avoid using System.outs / printfs directly and instead use (even a custom) logging class. That at least gives you a centralized kill-switch for all the loggings (minus the call costs in Java).
It is also useful to have that log class have info/warn/error/caveat, etc.
I would be careful about error levels, user ids, metadata, etc. since people don't always add them.
Also, one of the most common problems that I've seen is that people put temporary printfs in the code as they debug something, and then forget where they put them. I use a tool that tracks everything that I do so I can quickly identify all my recent edits since an abstract checkpoint and delete them. In most cases, however, you may want to pose special rules on debug code that can be checked into your source control.
In VB6 you've got
Debug.Print
which sends output to a window in the IDE. It's bearable for small projects. VB6 also has
#If <some var set in the project properties>
'debugging code
#End If
I have a logging class which I declare at the top with
Dim Trc as Std.Traces
and use in various places (often inside #If/#End If blocks)
Trc.Tracing = True
Trc.Tracefile = "c:\temp\app.log"
Trc.Trace 'with no argument stores date stamp
Trc.Trace "Var=" & var
Yes it does get messy, and yes I wish there was a better way.
We routinely are beginning to use a static class that we write trace messages to. It is very basic and still requires a call from the executing method, but it serves our purpose.
In the .NET world, there is already a fair amount of built in trace information available, so we do not need to worry about which methods are called or what the execution time is. These are more for specific events which occur in the execution of the code.
If your language does not support, through its tracing constructs, categorization of messages, it should be something that you add to your tracing code. Something to the effect that will identify different levels of importance and/or functional areas is a great start.
Just avoid instrumenting your code by modifying it. Learn to use a debugger. Make logging and error handling easy. Have a look at Aspect Oriented Programming
Debugging/Logging code can indeed be intrusive. In our C++ projects, we wrap common debug/log code in macros - very much like asserts. I often find that logging is most usefull in the lower level components so it doesn't have to go everywhere.
There is a lot in the other answers to both agree and disagree with :) Having debug/logging code can be a very valuable tool for troubleshooting problems. In Windows, there are many techniques - the two major ones are:
Extensive use of checked (DBG) build asserts and lots of testing on DBG builds.
the use of ETW in what we call 'fre' or 'retail' builds.
Checked builds (what most ohter call DEBUG builds) are very helpfull for us as well. We run all our tests on both 'fre' and 'chk' builds (on x86 and AMD64 as well, all serever stuff runs on Itanium too...). Some people even self host (dogfood) on checked builds. This does two things
Finds lots of bugs that woldn't be found otherwise.
Quickly elimintes noisy or unnessary asserts.
In Windows, we use Event Tracing for Windows (ETW) extensively. ETW is an efficient static logging mechanism. The NT kernel and many components are very well instrumented. ETW has a lot of advantages:
Any ETW event provider can be dynamically enabled/disabled at run time - no rebooting or process restarts required. Most ETW providers provide granular control over individual events, or groups of events.
Events from any provider (most importantly the kernel) can be merged into a single trace so all events can be correlated.
A merged trace can be copied off the box and fully processed - with symbols.
The NT kernel sample pofile interrupt can generate an ETW event - this yeilds a very light weight sample profiler that can be used any time
On Vista and Windows Server 2008, logging an event is lock free and fully multi-core aware - a thread on each processor can independently log events with no synchronization needed between them.
This is hugly valuable for us, and can be for your Windows code as well - ETW is usuable by any component - including user mode, drivers and other kernel components.
One thing we often do is write a streaming ETW consumer. Instead of putting printfs in the code, I just put ETW events at intersting places. When my componetn is running, I can then just run my ETW watcher at any time - the watcher receivs the events and displays them, conts them, or does other interesting things with them.
I very much respectfully disagree with tvanfosson. Even the best code can benefit from well implemented logging. Well implimented static run-time logging can make finding many problems straight forward - without it, you have zero visiblilty into what is happening in your component. You can look at inputs, outputs and guess -that's about it.
They key here is the term 'well implimented'. Instrumentation must be in the right place. Like any thing else, this requries some thought and planning. If it is not in helpfull/intersting places, then it will not help you find problems in a a development, testing, or deployed scenario. You can also have too much instrumeation causing perf problems when it is on - or even off!
Of course, different software products or componetns will have different needs. Some things may need very little instrumenation. But a widely depolyed or critical component can greatly benefit from weill egineered instrumeantion.
Here is a scenario for you (note, this very well may not apply to you...:) ). Lets say you have a line-of-business app deployed on every desktop in your company - hundreds or thousands of users. What do you do when someone has a pbolem? Do yo stop by their office and hookup a debugger? If so, how do you know what version they have? Where do you get the right symbols? How do you get the debuger on their system? What if it only happens once every few hours or days? Are you going to let the system run with the debugger connected all that time?
As you can imagine - hooking up debugger in this scenario is disruptive.
If your component is instrumented with ETW, then you could ask your user to simply turn on tracing; continue to do his/her work; then hit the "WTF" button when the problem happens. Even better: your app may have be able to self log - detecting problems at run time and turning on logging auto-magicaly. It could even send you ETW files when problems occured.
These are just simple exmples - logging can be handled many different ways. My key recomendation here is to think about how loging might be able to help you find, debug, and fix problems in your componetns at dev time, test time, and after they are deployed.
I was burnt by the same issue in about every project I've been involved with, so now I have this habit that involves extensive use of logging libraries (whatever the language/platform provides) from the start. Any Log4X port is fine for me.
Building yourself some proper debug tools can be extremely valuable. For example in a 3D environment, you might have an option to display the octree, or to render planned AI paths, or to draw waypoints that are normally invisible. You'd probably also want some on-screen display to aid with profiling too: the current framerate, count of polygons on screen, texture memory usage, and so on.
Although this takes some time and effort to do, in the long run it can save you a lot of time and frustration.