Search Results

Search found 5454 results on 219 pages for 'soa purge instances dehyd'.

Page 197/219 | < Previous Page | 193 194 195 196 197 198 199 200 201 202 203 204  | Next Page >

  • Win32_Product InstallLocation (error)

    - by andrew
    in C#, i'm trying to get some properties from the instances of Win32_Product, but i seem to have an error saying "Object reference not set to an instance of an object." here's the code: class Package { public string productName; public string installDate; public string installLocation; } class InstalledPackages { public static List<Package> get() { List<Package> packages = new List<Package>(); string query = "SELECT * FROM Win32_Product"; ManagementScope oMs = new ManagementScope(); ObjectQuery oQuery = new ObjectQuery(query); ManagementObjectSearcher oSearcher = new ManagementObjectSearcher(oMs, oQuery); ManagementObjectCollection installedPackages = oSearcher.Get(); foreach (ManagementObject package in installedPackages) { Package p = new Package(); p.productName = package["Name"].ToString(); p.installLocation = package["InstallLocation"].ToString(); p.installDate = package["InstallDate"].ToString(); packages.Add(p); } return packages; } } the exception appears when it gets to p.installLocation = package["InstallLocation"].ToString(); also, i get one if i try to do p.installLocation = package["InstallDate2"].ToString(); if i'm asking for InstallDate it works. (i'm using Windows 7 Ultimate x64)

    Read the article

  • Management API - The request body XML was invalid or not correctly specified

    - by maartenba
    Cross posting from http://social.msdn.microsoft.com/Forums/en-US/windowsazuretroubleshooting/thread/31b6aedc-c069-4e32-8e8f-2ff4b7c30793 I'm getting this error on changing configuration through the service management API: The request body XML was invalid or not correctly specified The request body payload: <?xml version="1.0" encoding="utf-8"?> <ChangeConfiguration xmlns="http://schemas.microsoft.com/windowsazu re"> <Configuration>PD94bWwgdmVyc2lvbj0iMS4wIj8+CjxTZXJ2aWNlQ29uZmlndX JhdGlvbiB4bWxuczp4c2k9Imh0dHA6Ly93d3cudzMub3JnLzIwMDEvWE1MU2NoZW1hLWluc3RhbmNlIi B4bWxuczp4c2Q9Imh0dHA6Ly93d3cudzMub3JnLzIwMDEvWE1MU2NoZW1hIiB4bWxucz0iaHR0cDovL3 NjaGVtYXMubWljcm9zb2Z0LmNvbS9TZXJ2aWNlSG9zdGluZy8yMDA4LzEwL1NlcnZpY2VDb25maWd1cm F0aW9uIiBzZXJ2aWNlTmFtZT0iIiBvc0ZhbWlseT0iMSIgb3NWZXJzaW9uPSIqIj4KICA8Um9sZSBuYW 1lPSJXZWJSb2xlMSI+CiAgICA8Q29uZmlndXJhdGlvblNldHRpbmdzPgogICAgICA8U2V0dGluZyBuYW 1lPSJNaWNyb3NvZnQuV2luZG93c0F6dXJlLlBsdWdpbnMuRGlhZ25vc3RpY3MuQ29ubmVjdGlvblN0cm luZyIgdmFsdWU9IlVzZURldmVsb3BtZW50U3RvcmFnZT10cnVlIi8+CiAgICA8L0NvbmZpZ3VyYXRpb2 5TZXR0aW5ncz4KICAgIDxJbnN0YW5jZXMgY291bnQ9IjIiLz4KICAgIDxDZXJ0aWZpY2F0ZXMvPgogID wvUm9sZT4KPC9TZXJ2aWNlQ29uZmlndXJhdGlvbj4K</Configuration> </ChangeConfiguration> I'm passing it the following configuration: $configuration = '<?xml version="1.0" encoding="utf-8"?> <ServiceConfiguration xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xmlns:xsd="http://www.w3.org/2001/XMLSchema" xmlns="http://schemas.microsoft.com/ServiceHosting/2008/10/ServiceConfiguration" serviceName="" osFamily="1" osVersion="*"> <Role name="WebRole1"> <ConfigurationSettings> <Setting name="Microsoft.WindowsAzure.Plugins.Diagnostics.ConnectionString" value="UseDevelopmentStorage=true"/> </ConfigurationSettings> <Instances count="2"/> <Certificates/> </Role> </ServiceConfiguration>'; Does anyone know why this error occurs? I suspect it has something to do with encoding but not sure.

    Read the article

  • Parsing an XML string containing "&#x20;" (which must be preserved)

    - by Zoodor
    I have code that is passed a string containing XML. This XML may contain one or more instances of &#x20; (an entity reference for the blank space character). I have a requirement that these references should not be resolved (i.e. they should not be replaced with an actual space character). Is there any way for me to achieve this? Basically, given a string containing the XML: <pattern value="[A-Z0-9&#x20;]" /> I do not want it to be converted to: <pattern value="[A-Z0-9 ]" /> (What I am actually trying to achieve is to simply take an XML string and write it to a "pretty-printed" file. This is having the side-effect of resolving occurrences of &#x20; in the string to a single space character, which need to be preserved. The reason for this requirement is that the written XML document must conform to an externally-defined specification.) I have tried creating a sub-class of XmlTextReader to read from the XML string and overriding the ResolveEntity() method, but this isn't called. I have also tried assigning a custom XmlResolver.

    Read the article

  • Setting CSS attributes on Change using jQuery

    - by Nick B
    I want to change css visibility and display attributes using jQuery on click when the state of another div's visibility attribute changes. (Many apologies for the obfuscated markup, but needing to manipulate someone else's construction): There are four instances of [data-label="Checkbox"] [data-label="Checked"] in this page. I want to set [data-label="trash"] and [data-label="Sort Options"] to visibility: visible; display: [empty value] when any of the [data-label="Checkbox"] [data-label="Checked"]'s attributes changes to 'visibility', 'visible'. Else, if none of [data-label="Checkbox"] [data-label="Checked"]'s have the attribute 'visibility', 'visible', I want to set [data-label="trash"] and [data-label="Sort Options"] back to their initial states: display: none; visibility: hidden;. Here's the markup: <div data-label="Sort Options" style="display: none; visibility: hidden;"> <div data-label="trash" style="display: none; visibility: hidden;"></div> </div> <div data-label="Checkbox"> <div data-label="Unchecked"></div> <div data-label="Checked" style="display: none; visibility: hidden;"></div> </div> Here is what I have tried unsuccessfully: $('[data-label="Checkbox"]').click(function() { if ('[data-label="Checkbox"] [data-label="Checked"]').css('visibility', 'visible') { $('[data-label="trash"], [data-label="Sort Options"]').css({'display': '', 'visibility': 'visible'}); } else { $('[data-label="trash"], [data-label="Sort Options"]').css({'display': 'none', 'visibility': 'hidden'}); } }); Any help would be greatly appreciated! Thanks

    Read the article

  • C++ Instantiate class and add name to array/vector then cycle through update functions from the array

    - by SD42
    I have a bunch of classes in a basic (and badly coded!) game engine that are currently instantiated in the code and have to have their individual update functions called. What I want to do is be able to create an instance of a class, pass the name to an array, and then subsequently cycle through the array to call the update functions of each class. I'm unsure as to whether or not this is an impossible or spectacularly stupid way of trying to manage objects, so please tell me if it is. So currently I might have manually instantiated Enemy class a couple of times: Enemy enem1; Enemy enem2; I then have to update them manually in the game loop: enem1.update(); enem2.update(); What method should I use to be able to spawn and destroy instances of the enemy class during gametime? Is it possible to populate an array with instantiated names and then do something like (and i'm aware this doesn't work); array[x].update(); Then iterate through the names in the array? Anything that even points me in the right direction would be greatly appreciated!

    Read the article

  • Built in background-scheduling system in .NET?

    - by Lasse V. Karlsen
    I ask though I doubt there is any such system. Basically I need to schedule tasks to execute at some point in the future (usually no more than a few seconds or possibly minutes from now), and have some way of cancelling that request unless too late. Ie. code that would look like this: var x = Scheduler.Schedule(() => SomethingSomething(), TimeSpan.FromSeconds(5)); ... x.Dispose(); // cancels the request Is there any such system in .NET? Is there anything in TPL that can help me? I need to run such future-actions from various instances in a system here, and would rather avoid each such class instance to have its own thread and deal with this. Also note that I don't want this (or similar, for instance through Tasks): new Thread(new ThreadStart(() => { Thread.Sleep(5000); SomethingSomething(); })).Start(); There will potentially be a few such tasks to execute, they don't need to be executed in any particular order, except for close to their deadline, and it isn't vital that they have anything like a realtime performance concept. I just want to avoid spinning up a separate thread for each such action.

    Read the article

  • Using a single texture image unit with multiple sampler uniforms

    - by bcrist
    I am writing a batching system which tracks currently bound textures in order to avoid unnecessary glBindTexture() calls. I'm not sure if I need to keep track of which textures have already been used by a particular batch so that if a texture is used twice, it will be bound to a different TIU for the second sampler which requires it. Is it acceptable for an OpenGL application to use the same texture image unit for multiple samplers within the same shader stage? What about samplers in different shader stages? For example: Fragment shader: ... uniform sampler2D samp1; uniform sampler2D samp2; void main() { ... } Main program: ... glActiveTexture(GL_TEXTURE0); glBindTexture(GL_TEXTURE_2D, tex_id); glUniform1i(samp1_location, 0); glUniform1i(samp2_location, 0); ... I don't see any reason why this shouldn't work, but what about if the shader program also included a vertex shader like this: Vertex shader: ... uniform sampler2D samp1; void main() { ... } In this case, OpenGL is supposed to treat both instances of samp1 as the same variable, and exposes a single location for them. Therefore, the same texture unit is being used in the vertex and fragment shaders. I have read that using the same texture in two different shader stages counts doubly against GL_MAX_COMBINED_TEXTURE_IMAGE_UNITS but this would seem to contradict that. In a quick test on my hardware (HD 6870), all of the following scenarios worked as expected: 1 TIU used for 2 sampler uniforms in same shader stage 1 TIU used for 1 sampler uniform which is used in 2 shader stages 1 TIU used for 2 sampler uniforms, each occurring in a different stage. However, I don't know if this is behavior that I should expect on all hardware/drivers, or if there are performance implications.

    Read the article

  • NSPredicate (Core Data fetch) to filter on an attribute value being present in a supplied set (list)

    - by starbaseweb
    I'm trying to create a fetch predicate that is the analog to the SQL "IN" statement, and the syntax to do so with NSPredicate escapes me. Here's what I have so far (the relevant excerpt from my fetching routine): NSFetchRequest *request = [[[NSFetchRequest alloc] init] autorelease]; NSEntityDescription *entity = [NSEntityDescription entityForName: @"BodyPartCategory" inManagedObjectContext:_context]; [request setEntity:entity]; NSPredicate *predicate = [NSPredicate predicateWithFormat:@"(name IN %@)", [RPBodyPartCategory defaultBodyPartCategoryNames]]; [request setPredicate:predicate]; The entity "BodyPartCategory" has a string attribute "name". I have a list of names (just NSString objects) in an NSArray as returned by: [RPBodyPartCategory defaultBodyPartCategoryNames] So let's say that array has string such as {@"Liver", @"Kidney", @"Thyroid"} ... etc. I want to fetch all 'BodyPartCategory' instances whose name attribute matches one of the strings in the set provided (technically NSArray but I can make it an NSSet). In SQL, this would be something like: SELECT * FROM BodyPartCategories WHERE name IN ('Liver', 'Kidney', 'Thyroid') I've gone through various portions of the Predicate Programming Guide, but I don't see this simple use case covered. Pointers/help much appreciated!

    Read the article

  • How to get correct children ids using fields_for "parents[]", parent do |f| using f.fields_for :children, child ?

    - by Anatortoise House
    I'm editing multiple instances of a parent model in an index view in one form, as in Railscasts #198. Each parent has_many :children and accepts_nested_attributes_for :children, as in Railscasts #196 and #197 <%= form_tag %> <% for parent in @parents %> <%= fields_for "parents[]", parent do |f| <%= f.text_field :job %> <%= f.fields_for :children do |cf| %> <% cf.text_field :chore %> <% end %> <% end %> <% end %> <% end %> Given parent.id==1 f.text_field :job correctly generates <input id="parents_1_job" type="text" value="coding" size="30" name="parents[1][job]"> But cf.text_field :chore generates ids and names that don't have the parent index. id="parents_children_attributes_0_chore" name="parents[children_attributes][0][chore]" If I try passing the specific child object to f.fields_for like this: <% for child in parent.children %> <%= f.fields_for :children, child do |cf| %> <%= cf.text_field :chore %> <% end %> <% end %> I get the same. If I change the method from :children to "[]children" I get id="parents_1___children_chore" which gets the right parent_index but doesn't provide an array slot for the child index. "[]children[]" isn't right either: id="parents_1__children_3_chore" as I was expecting attributes_0_chore instead of 3_chore. Do I need to directly modify an attribute of the FormBuilder object, or subclass FormBuilder to make this work, or is there a syntax that fits this situation? Thanks for any thoughts.

    Read the article

  • ASP.NET configuration inheritance

    - by NowYouHaveTwoProblems
    I have an ASP.NET application that defines a custom configuration section in web.config. Recently I had a customer who wanted to deploy two instances of the application (for testing in addition to an existing production application). The configuration chosen by the customer was: foo.com - production application foo.com/Testing - test application In this case, the ASP.NET configuration engine decided to apply the settings at foo.com/web.config to foo.com/Testing/web.config. Thankfully this caused a configuration error because the section was redefined at the second level rather than giving the false impression that the two web applications were isolated. What I would like to do is to specify that my configuration section is not inherited and must be re-defined for any web application that requires it but I haven't been able to find a way to do this. My web.config ends up something like this <configuration> <configSections> <section name="MyApp" type="MyApp.ConfigurationSection"/> </configSections> <MyApp setting="value" /> <NestedSettingCollection> <Item key="SomeKey" value="SomeValue" /> <Item key="SomeOtherKey" value="SomeOtherValue" /> </NestedSettingCollection> </MyApp> </configuration>

    Read the article

  • Does Entity Framework saves related classes automatically?

    - by herbatnic
    Let's assume that we have such classes public class A{ string someField { get; set; } public virtual B B {get; set; } } public class B { int someIntField {get; set; } [ForeignKey("Id")] [Required] public virtual A A { get; set; } } In code I create new instances for both of them and making relation like: A a = new A () { someField = "abcd"}; B b = new B () { someIntField = 42 }; A.B = b; B.A = a; Should I using DBContext to save both classes like that: using (var db = new myDbContext()) { myDbContext.As.Add(A); myDbContext.Bs.Add(B); myDBContext.SaveChanges(); } Or saving it like that: using (var db = new myDbContext()) { myDbContext.As.Add(A); myDbContext.SaveChanges(); } is enough to store related objects into database?

    Read the article

  • How to structure Javascript programs in complex web applications?

    - by mixedpickles
    Hi there. I have a problem, which is not easily described. I'm writing a web application that makes heavy usage of jquery and ajax calls. Now I don't have experience in designing the architecture for javascript programms, but I realize that my program has not a good structure. I think I have to many identifiers referring to the same (at least more or less) thing. Let's have an exemplary look at an arbitrary UI widget: The eventhandlers use DOM elements as parameters. The DOM element represents a widget in the browser. A lot of times I use jQuery objects (I think they are basically a wrapper around DOM elements) to do something with the widget. Sometimes they are used transiently, sometimes they are stored in a variable for later purposes. The ajax function calls use strings identifiers for these widgets. They are processed server side. Beside that I have a widget class whose instances represents a widget. It is instantiated through the new operator. Now I have somehow four different object identifiers for the same thing, which needs to be kept in sync until the page is loaded anew. This seems not to be a good thing. Any advice?

    Read the article

  • How can I bind the same dependency to many dependents in Ninject?

    - by Mike Bantegui
    Let's I have three interfaces: IFoo, IBar, IBaz. I also have the classes Foo, Bar, and Baz that are the respective implementations. In the implementations, each depends on the interface IContainer. So for the Foo (and similarly for Bar and Baz) the implementation might read: class Foo : IFoo { private readonly IDependency Dependency; public Foo(IDependency dependency) { Dependency = dependency; } public void Execute() { Console.WriteLine("I'm using {0}", Dependency.Name); } } Let's furthermore say I have a class Container which happens to contain instances of the IFoo, IBar and IBaz: class Container : IContainer { private readonly IFoo _Foo; private readonly IBar _Bar; private readonly IBaz _Baz; public Container(IFoo foo, IBar bar, IBaz baz) { _Foo = foo; _Bar = bar; _Baz = baz; } } In this scenario, I would like the implementation class Container to bind against IContainer with the constraint that the IDependency that gets injected into IFoo, IBar, and IBaz be the same for all three. In the manual way, I might implement it as: IDependency dependency = new Dependency(); IFoo foo = new Foo(dependency); IBar bar = new Bar(dependency); IBaz baz = new Baz(dependency); IContainer container = new Container(foo, bar, baz); How can I achieve this within Ninject? Note: I am not asking how to do nested dependencies. My question is how I can guarantee that a given dependency is the same among a collection of objects within a materialized service. To be extremely explicit, I understand that Ninject in it's standard form will generate code that is equivalent to the following: IContainer container = new Container(new Foo(new Dependency()), new Bar(new Dependency()), new Baz(new Dependency())); I would not like that behavior.

    Read the article

  • MySQL Table Loop using PHP

    - by JM4
    I have an online form which collects consumer data and stores in a dedicated MySQL database. In some instances, data is passed in the URL under the "RefID" variable which is also stored in the database and attached to each registration. I use the 'mysql_num_rows ($result)' to fetch all agent details on another page but this only returns ALL available details. My goal is as follows: GOAL I want to create an HTML table in which rows are automatically generated based on the list of all registrations on my site. A new row is created IF and ONLY IF a unique RefID is present on that particular record. In the event the field is NULL, it is reported on a single line. In short, the HTML table could look something like this: RefID - Number of Enrollments abc123 - 10 baseball - 11 twonk - 7 NULL - 33 Where abc123 is a particular RefID and 10 is the number of times that RefID appears in the DB. If a new registration comes in with RefID = "horses", a new row is created, showing "horses - 1". The HTML table will be viewable by account administrators needing to see the number of enrollments for a particular RefID (which they won't know ahead of time). Anybody have any suggestions?

    Read the article

  • Initializing ExportFactory using MEF

    - by Riz
    Scenario Application has multiple parts. Each part is in separate dll and implements interface IFoo All such dlls are present in same directory (plugins) The application can instantiate multiple instances of each part Below is the code snippet for the interfaces, part(export) and the import. The problem I am running into is, the "factories" object is initialized with empty list. However, if I try container.Resolve(typeof(IEnumerable< IFoo )) I do get object with the part. But that doesn't serve my purpose (point 4). Can anyone point what I am doing wrong here? public interface IFoo { string Name { get; } } public interface IFooMeta { string CompType { get; } } Implementation of IFoo in separate Dll [ExportMetadata("CompType", "Foo1")] [Export(typeof(IFoo), RequiredCreationPolicy = CreationPolicy.NonShared))] public class Foo1 : IFoo { public string Name { get { return this.GetType().ToString(); } } } Main application that loads all the parts and instantiate them as needed class PartsManager { [ImportMany] private IEnumerable<ExportFactory<IFoo, IFooMeta>> factories; public PartsManager() { IContainer container = ConstructContainer(); factories = (IEnumerable<ExportFactory<IFoo, IFooMeta>>) container.Resolve(typeof(IEnumerable<ExportFactory<IFoo, IFooMeta>>)); } private static IContainer ConstructContainer() { var catalog = new DirectoryCatalog(@"C:\plugins\"); var builder = new ContainerBuilder(); builder.RegisterComposablePartCatalog(catalog); return builder.Build(); } public IFoo GetPart(string compType) { var matchingFactory = factories.FirstOrDefault( x => x.Metadata.CompType == compType); if (factories == null) { return null; } else { IFoo foo = matchingFactory.CreateExport().Value; return foo; } } }

    Read the article

  • grails services :: multiple projects

    - by naveen
    PROBLEM : I have multiple grails projects (lets say appA, appB and appC) : services to be precise I want to run them in a single grails-app.. probably a war deployment, how can i do this? REQUIREMENTS : I want this to be a single app since i am deploying it on cloud and i don't have enough memory to hold all these service instances individually. The reason for multiple grails project is scalability. So that if later on i want to run 10 instance of appA, 3 instance of appB, and 1 instance of aapC; i should be able to do that. EDIT : Can i use something like 0mq, will that be helpful in keeping the services separated from each other. How will i package my service? And reading the docs of 0mq seems that it can work with both inprocess and external process. Will async grails requests on HTTP work with 0mq in process/ external mq calls. Haven't used 0mq, but from the initial doc it seems to work. Need some experience calls in this scenario. Are there any other alternatives or mq alternatives?

    Read the article

  • Prevent cached objects to end up in the database with Entity Framework

    - by Dirk Boer
    We have an ASP.NET project with Entity Framework and SQL Azure. A big part of our data only needs to be updated a few times a day, other data is very volatile. The data that barely changes we cache in memory at startup, detach from the context and than use it mainly for reading, drastically lowering the amount of database requests we have to do. The volatile data is requested everytime by a DbContext per Http request. When we do an update to the cached data, we send a message to all instances to catch a fresh version of all the data from the SQL server. So far, so good. Until we introduced a bug that linked one of these 'cached' objects to the 'volatile' data, and did a SaveChanges. Well, that was quite a mess. The whole data tree was added again and again by every update, corrupting the whole database with a whole lot of duplicated data. As a complete hack I added a completely arbitrary column with a UniqueConstraint and some gibberish data on one of the root tables; hopefully failing the SaveChanges() next time we introduce such a bug because it will violate the Unique Constraint. But it is of course hacky, and I'm still pretty scared ;P Are there any better ways to prevent whole tree's of cached objects ending up in the database? More information Project is ASP.NET MVC I cache this data, because it is mainly read only, and this saves a tons of extra database calls per http request

    Read the article

  • What does this Java generics paradigm do and what is it called?

    - by Tom
    I'm looking at some Java classes that have the following form: public abstract class A <E extends A<E>> implements Comparable <E> { public final int compareTo( E other ) { // etc } } public class B extends A <B> { // etc } public class C extends A <C> { // etc } My usage of "Comparable" here is just to illustrate a possible use of the generic parameter "E". Does this usage of generics/inheritance have a name? What is it used for? My impression is that this allows the abstract class to provide a common implementation of a method (such as compareTo) without having to provide it in the subclasses. However, in this example, unlike an inherited method it would restrict subclasses to invoking compareTo on other instances of the same subclass, rather than any "A" subclass. Does this sound right? Anyway, just curious if any gurus out there have seen this before and know what it does. Thanks!

    Read the article

  • Generating an identifier for objects so that they can be added to a hashtable I have created

    - by dukenukem
    I have a hashtable base class and I am creating different type of hashtable by deriving from it. I only allow it to accept objects that implement my IHashable interface.For example - class LinearProbingHashTable<T> : HashTableBase<T> where T: IHashable { ... ... ... } interface IHashable { /** * Every IHashable implementation should provide an indentfying value for use in generating a hash key. */ int getIdentifier(); } class Car : IHashable { public String Make { get; set; } public String Model { get; set; } public String Color { get; set; } public int Year { get; set; } public int getIdentifier() { /// ??? } } Can anyone suggest a good method for generating an identifier for the car that can be used by the hash function to place it in the hash table? I am actually really looking for a general purpose solution to generating an id for any given class. I would like to have a base class for all classes, HashableObject, that implements IHashable and its getIdentifier method. So then I could just derive from HashableObject which would automatically provide an identifier for any instances. Which means I wouldn't have to write a different getIdentifier method for every object I add to the hashtable. public class HashableObject : IHashable { public int getIdentifier() { // Looking for code here that would generate an id for any object... } } public class Dog : HashableObject { // Dont need to implement getIdentifier because the parent class does it for me }

    Read the article

  • JS best practice for member functions

    - by MickMalone1983
    I'm writing a little mobile games library, and I'm not sure the best practice for declaring member functions of instantiated function objects. For instance, I might create a simple object with one property, and a method to print it: function Foo(id){ this.id = id; this.print = function(){ console.log(this.id); }; }; However, a function which does not need access to 'private' members of the function does not need to be declared in the function at all. I could equally have written: function print(){ console.log(this.id); }; function Foo(id){ this.id = id; this.print = print; }; When the function is invoked through an instance of Foo, the instance becomes the context for this, so the output is the same in either case. I'm not entirely sure how memory is allocated with JS, and I can't find anything that I can understand about something this specific, but it seems to me that with the first example all members of Foo, including the print function, are duplicated each time it is instantiated - but with the second, it just gets a pointer to one, pre-declared function, which would save any more memory having to be allocated as more instances of Foo are created. Am I correct, and if I am, is there any memory/performance benefit to doing this?

    Read the article

  • setting up bind to work with nsupdate (SERVFAIL)

    - by funny_ha_ha
    I'm trying to update my DNS-Server dynamically using nsupdate. Prerequisite I'm using Debian 6 on my DNS-Server and Debian 4 on my client. I created a public/private key pair using: dnssec-keygen -C -a HMAC-MD5 -b 512 -n USER sub.example.com. I then edited my named.conf.local to contain my public key and the new zone i wish to update. It now looks like this (note: I also tried allow-update { any; }; without success): zone "example.com" { type master; file "/etc/bind/primary/example.com"; notify yes; allow-update { none; }; allow-query { any; }; }; zone "sub.example.com" { type master; file "/etc/bind/primary/sub.example.com"; notify yes; allow-update { key "sub.example.com."; }; allow-query { any; }; }; key sub.example.com. { algorithm HMAC-MD5; secret "xxxx xxxx"; }; Next, I copied the private key file (key.private) to another server I want to update the zone from. I also created a textfile (update) on this server which contained the update information (note: I tried toying around with this stuff too. no success): server example.com zone sub.example.com update add sub.example.com. 86400 A 10.10.10.1 show send Now I'm trying to update the zone using: nsupdate -k key.private -v update The Problem Said command gives me the following output: Outgoing update query: ;; ->>HEADER<<- opcode: UPDATE, status: NOERROR, id: 0 ;; flags: ; ZONE: 0, PREREQ: 0, UPDATE: 0, ADDITIONAL: 0 ;; ZONE SECTION: ;sub.example.com. IN SOA ;; UPDATE SECTION: sub.example.com. 86400 IN A 10.10.10.1 update failed: SERVFAIL named debug Level 3 gives me the following information when I issue the nsupdate command on the remote server (note: I obfuscated the client IP): 06-Aug-2012 14:51:33.977 client X.X.X.X#33182: new TCP connection 06-Aug-2012 14:51:33.977 client X.X.X.X#33182: replace 06-Aug-2012 14:51:33.978 clientmgr @0x2ada3c7ee760: createclients 06-Aug-2012 14:51:33.978 clientmgr @0x2ada3c7ee760: recycle 06-Aug-2012 14:51:33.978 client @0x2ada475f1120: accept 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: read 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: TCP request 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: request has valid signature 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: recursion not available 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: update 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: send 06-Aug-2012 14:51:33.978 client X.X.X.X#33182: sendto 06-Aug-2012 14:51:33.979 client X.X.X.X#33182: senddone 06-Aug-2012 14:51:33.979 client X.X.X.X#33182: next 06-Aug-2012 14:51:33.979 client X.X.X.X#33182: endrequest 06-Aug-2012 14:51:33.979 client X.X.X.X#33182: read 06-Aug-2012 14:51:33.986 client X.X.X.X#33182: next 06-Aug-2012 14:51:33.986 client X.X.X.X#33182: request failed: end of file 06-Aug-2012 14:51:33.986 client X.X.X.X#33182: endrequest 06-Aug-2012 14:51:33.986 client X.X.X.X#33182: closetcp But it doesn't do anything. The zone isn't updated, nor does my nsupdate change anything. I'm not sure if the file /etc/bind/primary/sub.example.com should exist prior to the first update or not. I tried it without the file, with an empty file and with a pre-configured zone file. Without success. The sparse information I found on the net pointed me towards file and folder permissions regarding the bind working directory, so I changed the permissions of both /etc/bind and /var/cache/bind (which is the home dir of my "bind" user). I'm not a 100% sure if the permissions are correct.. but it looks good to me: ls -lah /var/cache/bind/ total 224K drwxrwxr-x 2 bind bind 4.0K Aug 6 03:13 . drwxr-xr-x 12 root root 4.0K Jul 21 11:27 .. -rw-r--r-- 1 bind bind 211K Aug 6 03:21 named.run ls -lah /etc/bind/ total 72K drwxr-sr-x 3 bind bind 4.0K Aug 6 14:41 . drwxr-xr-x 87 root root 4.0K Jul 30 01:24 .. -rw------- 1 bind bind 125 Aug 6 02:54 key.public -rw------- 1 bind bind 156 Aug 6 02:54 key.private -rw-r--r-- 1 bind bind 2.5K Aug 6 03:07 bind.keys -rw-r--r-- 1 bind bind 237 Aug 6 03:07 db.0 -rw-r--r-- 1 bind bind 271 Aug 6 03:07 db.127 -rw-r--r-- 1 bind bind 237 Aug 6 03:07 db.255 -rw-r--r-- 1 bind bind 353 Aug 6 03:07 db.empty -rw-r--r-- 1 bind bind 270 Aug 6 03:07 db.local -rw-r--r-- 1 bind bind 3.0K Aug 6 03:07 db.root -rw-r--r-- 1 bind bind 493 Aug 6 03:32 named.conf -rw-r--r-- 1 bind bind 490 Aug 6 03:07 named.conf.default-zones -rw-r--r-- 1 bind bind 1.2K Aug 6 14:18 named.conf.local -rw-r--r-- 1 bind bind 666 Jul 29 22:51 named.conf.options drwxr-sr-x 2 bind bind 4.0K Aug 6 03:57 primary/ -rw-r----- 1 root bind 77 Mar 19 02:57 rndc.key -rw-r--r-- 1 bind bind 1.3K Aug 6 03:07 zones.rfc1918 ls -lah /etc/bind/primary/ total 20K drwxr-sr-x 2 bind bind 4.0K Aug 6 03:57 . drwxr-sr-x 3 bind bind 4.0K Aug 6 14:41 .. -rw-r--r-- 1 bind bind 356 Jul 30 00:45 example.com

    Read the article

  • How to properly add .NET assemblies to Powershell session?

    - by amandion
    I have a .NET assembly (a dll) which is an API to backup software we use here. It contains some properties and methods I would like to take advantage of in my Powershell script(s). However, I am running into a lot of issues with first loading the assembly, then using any of the types once the assembly is loaded. The complete file path is: C:\rnd\CloudBerry.Backup.API.dll In Powershell I use: $dllpath = "C:\rnd\CloudBerry.Backup.API.dll" Add-Type -Path $dllpath I get the error below: Add-Type : Unable to load one or more of the requested types. Retrieve the LoaderExceptions property for more information. At line:1 char:9 + Add-Type <<<< -Path $dllpath + CategoryInfo : NotSpecified: (:) [Add-Type], ReflectionTypeLoadException + FullyQualifiedErrorId : System.Reflection.ReflectionTypeLoadException,Microsoft.PowerShell.Commands.AddTypeComma ndAdd-Type : Unable to load one or more of the requested types. Retrieve the LoaderExceptions property for more information. Using the same cmdlet on another .NET assembly, DotNetZip, which has examples of using the same functionality on the site also does not work for me. I eventually find that I am seemingly able to load the assembly using reflection: [System.Reflection.Assembly]::LoadFrom($dllpath) Although I don't understand the difference between the methods Load, LoadFrom, or LoadFile that last method seems to work. However, I still seem to be unable to create instances or use objects. Each time I try, I get errors that describe that Powershell is unable to find any of the public types. I know the classes are there: $asm = [System.Reflection.Assembly]::LoadFrom($dllpath) $cbbtypes = $asm.GetExportedTypes() $cbbtypes | Get-Member -Static ---- start of excerpt ---- TypeName: CloudBerryLab.Backup.API.BackupProvider Name MemberType Definition ---- ---------- ---------- PlanChanged Event System.EventHandler`1[CloudBerryLab.Backup.API.Utils.ChangedEventArgs] PlanChanged(Sy... PlanRemoved Event System.EventHandler`1[CloudBerryLab.Backup.API.Utils.PlanRemoveEventArgs] PlanRemoved... CalculateFolderSize Method static long CalculateFolderSize() Equals Method static bool Equals(System.Object objA, System.Object objB) GetAccounts Method static CloudBerryLab.Backup.API.Account[], CloudBerry.Backup.API, Version=1.0.0.1, Cu... GetBackupPlans Method static CloudBerryLab.Backup.API.BackupPlan[], CloudBerry.Backup.API, Version=1.0.0.1,... ReferenceEquals Method static bool ReferenceEquals(System.Object objA, System.Object objB) SetProfilePath Method static System.Void SetProfilePath(string profilePath) ----end of excerpt---- Trying to use static methods fail, I don't know why!!! [CloudBerryLab.Backup.API.BackupProvider]::GetAccounts() Unable to find type [CloudBerryLab.Backup.API.BackupProvider]: make sure that the assembly containing this type is load ed. At line:1 char:42 + [CloudBerryLab.Backup.API.BackupProvider] <<<< ::GetAccounts() + CategoryInfo : InvalidOperation: (CloudBerryLab.Backup.API.BackupProvider:String) [], RuntimeException + FullyQualifiedErrorId : TypeNotFound Any guidance appreciated!!

    Read the article

  • "dig +trace fqdn" and "dig fqdn" do not give the same result on a LAN with windows DNS server, why?

    - by Sulliwane
    in my company LAN I have a Ubuntu 14.04 server running in Virtualbox (as guest) on a Windows 7 (the host) with network interface bridged (so the Ubuntu server belongs to the LAN, with its ip: 192.168.1.85). I have a website on this server: mywebsite.com The gateway for the LAN to the internet is 192.168.1.1 (Cisco 1841)--188.188.188.254 as public IP. There is a Windows 2008 server that acts as DNS server and DHCP server on the LAN. I added a Forward zone "mywebsite.com" with A record - 192.168.1.85. Outside the LAN, mywebsite.com has public Dns records that point on the Cisco 1841 public IP (188.188.188.254) Now when I ping mywebsite.com from the lan, I quickly get 192.168.1.85. But when I'm connecting through the browser on the clients, it's not always fast. So I'm wondering: Are my requests really/directly resolved and forwarded to 192.168.1.85, OR are they sent out of the LAN, and then forwarded back to the CISCO public 188.188.188.254:80 and NAT to the Ubuntu server before being served ??? To try to answer this question, I looked for tracking the DNS request from my linux client on the LAN: v@v-ss9:~$ dig mywebsite.com ; <<>> DiG 9.9.5-3-Ubuntu <<>> mywebsite.com ;; global options: +cmd ;; Got answer: ;; ->>HEADER<<- opcode: QUERY, status: NOERROR, id: 24850 ;; flags: qr aa rd ra; QUERY: 1, ANSWER: 1, AUTHORITY: 0, ADDITIONAL: 1 ;; OPT PSEUDOSECTION: ; EDNS: version: 0, flags:; udp: 4000 ;; QUESTION SECTION: ;mywebsite.com. IN A ;; ANSWER SECTION: mywebsite.com. 3600 IN A 192.168.1.85 ;; Query time: 1 msec ;; SERVER: 127.0.1.1#53(127.0.1.1) ;; WHEN: Fri Aug 22 09:50:16 CST 2014 ;; MSG SIZE rcvd: 66 This answer looks right: 192.168.1.85. But then look at this: v@v-ss9:~$ dig +trace mywebsite.com ; <<>> DiG 9.9.5-3-Ubuntu <<>> +trace mywebsite.com ;; global options: +cmd . 12955 IN NS h.gtld-servers.net. . 12955 IN NS g.gtld-servers.net. . 12955 IN NS m.gtld-servers.net. . 12955 IN NS i.gtld-servers.net. . 12955 IN NS l.gtld-servers.net. . 12955 IN NS k.gtld-servers.net. . 12955 IN NS j.gtld-servers.net. . 12955 IN NS d.gtld-servers.net. . 12955 IN NS b.gtld-servers.net. . 12955 IN NS c.gtld-servers.net. . 12955 IN NS a.gtld-servers.net. . 12955 IN NS e.gtld-servers.net. . 12955 IN NS f.gtld-servers.net. ;; Received 516 bytes from 127.0.1.1#53(127.0.1.1) in 18 ms mywebsite.com. 172800 IN NS ns3.rmi.fr. mywebsite.com. 172800 IN NS ns4.rmi.fr. CK0POJMG874LJREF7EFN8430QVIT8BSM.com. 86400 IN NSEC3 1 1 0 - CK0QFMDQRCSRU0651QLVA1JQB21IF7UR NS SOA RRSIG DNSKEY NSEC3PARAM CK0POJMG874LJREF7EFN8430QVIT8BSM.com. 86400 IN RRSIG NSEC3 8 2 86400 20140825045016 20140818034016 6122 com. Imq8K9xlvFXlB4IjUkdxOc5YHoTEhqSQUlRSJ9QCIhd9wzGpWJ54AfVf WJ0SUKThalpzqS0cXdLGtNmuYgqLfwUMjpUlT4c+zJyx7I4QMPLImQZh Ov0xy3mUr7dLlymAJYGs9dLI2IaheLvpKTBwaV1gAvo8QEkU8VRiJ7gW 9dk= U0PIA23FHMVPTKSDHC9PJ1BEA9SIB65R.com. 86400 IN NSEC3 1 1 0 - U0PL33R61V6TCCPBS1171PROP57ASRD9 NS DS RRSIG U0PIA23FHMVPTKSDHC9PJ1BEA9SIB65R.com. 86400 IN RRSIG NSEC3 8 2 86400 20140825043502 20140818032502 6122 com. qsC5sJbwklao+OedCHpcYo56aQaY0N+7peKmPu8szvjAQoJFRWyuDfAh Nw/gvHXEMzG7tYLriQGVfsiK8GZdPXyG4Ghe1MNN4jOZnSahkT5LjlqL 5QyGC0QiClRMPDAYjUOFGQDkjOJcJYvTNkEyXC2BEpfLI5SwCbYqwqg3 RkE= ;; Received 585 bytes from 192.41.162.30#53(l.gtld-servers.net) in 297 ms mywebsite.com. 86400 IN A 188.188.188.254 mywebsite.com. 86400 IN NS ns3.rmi.fr. mywebsite.com. 86400 IN NS ns4.rmi.fr. ;; Received 204 bytes from 212.51.161.18#53(ns3.rmi.fr) in 310 ms Here I get my CISCO public IP 188.188.188.254!!! Is it normal? How to know if my browser (from the LAN) is really directly communicating with 192.168.1.85 when using mywebsite.com? Thank you for your help.

    Read the article

  • Ubuntu 12.04: apt-get "failed to fetch"; apt is trying to fetch via old static IP

    - by gabe
    Sample error: W: Failed to fetch http://security.ubuntu.com/ubuntu/dists/precise-security/universe/i18n/Translation-en Unable to connect to 192.168.1.70:8118: Now this was working just fine until I changed the IP this morning. I have the server set to a static IP of 10.0.1.70 and for years it has been 192.168.1.70 - the IP apt-get is trying to use right now. I use privoxy and tor thus the 8118 port. Like I said it all worked until I changed the static IP from 192.168.1.70 to 10.0.1.70. I was forced to do so because of router issues. (Long and involved story, I didn't really want to change the IP because I know something like this would happen.) The setup for TOR/Privoxy requires that has you point Privoxy at TOR via 127.0.0.1:9050. Then point curl, etc to Privoxy via $HOME/.bashrc. Typically you would set the listen to IP for Privoxy to 127.0.0.1 but if you want it accessible to the rest of the LAN you set the IP to the server's LAN IP. Which I did a long time ago and was working fine until this morning. I have changed all instances of 192.168.1.70 to 10.0.1.70 in both /etc/privoxy/config and $HOME/.bashrc. What makes this really strange for me is that curl is working fine. I curl icanhazip.com and voila I get a new IP every 10 minutes or so. I curl CNN.com and I get the short but sweet permanently moved to www.cnn.com message I expect. Firefox works fine. Ping works fine. And I've tested all of this via Remote Desktop over my LAN. So the connection appears to be fine for everything except apt. I've also rebooted hoping that would clear 192.168.1.70 from apt. So the connection to the internet and DNS aren't an issue for these programs. And they are, as far as I can tell, using Privoxy/TOR just fine. The real irony here is that I've tried to open up Privoxy to go to Ubuntu's servers directly without going through TOR to speed up the downloads from Ubuntu (did this months ago). So somewhere that I have not been able to find, apt has stored the IP 192.168.1.70. And 192.168.1.70 is no longer valid. Thanks for the help

    Read the article

  • Issues with Server 2012 using DFSR running on Hyper-V 2012

    - by Bryan
    We have a number of Server 2012 systems, all of which run virtualised on Hyper-V 2012 server. We are having problems with two such virtual instances, both of which are used as file servers, whereby they occasionally stop responding to requests to serve files to clients. After logging on to the server, attempts to shut it down gracefully fail (no error, it just fails to acknowledge a shutdown request). Recovery is a case of power cycling the server(s) from the Hyper-V console. These two servers don't server a large number of users (one serves no more than 6 users, and the other serves around 20 users), they are in the same domain, but on different physical hardware (and at different sites). They don't lock up at the same time. They both use DFSR to replicate a fairly large amount of data between themselves (200GB) over ADSL connections, this is working fine, and we have been using DFSR to do this on the previous two generations of server OS we have used (Server 2008 R2 and Server 2003 - both of which were physical installs however). Today, when one of the servers crashed, I noticed an entry in the event log, which looked similar to the following: Log Name: Application Source: ESENT Date: 27/11/2012 10:25:55 Event ID: 533 Task Category: General Level: Warning Keywords: Classic User: N/A Computer: HAL-FS-01.example.com Description: DFSRs (1500) \\.\E:\System Volume Information\DFSR\database_C8CC_101_CC00_EC0E\ dfsr.db: A request to write to the file "\\.\E:\System Volume Information\ DFSR\database_C8CC_101_CC00_EC0E\fsr.log" at offset 4423680 (0x0000000000438000) for 4096 (0x00001000) bytes has not completed for 36 second(s). This problem is likely due to faulty hardware. Please contact your hardware vendor for further assistance diagnosing the problem. When the server started up again, I went to find the event log entry to investigate further and found that the event log entry was no longer there (I assume it was in memory but failed to write to disk before the server was powered off, for the reason mentioned in the message). I found the above message by searching back further in the event log. Both of these virtual servers have their E: volumes fully allocated as opposed to dynamically expanding, and there are no other issues on any of the other virtual servers (which include server 2012, server 2008 R2 and Ubuntu 12.04 x64). There are no signs of IO, memory or CPU starvation on the host systems. I've used performance counters on the affected virtual servers to monitor memory usage (including non paged pool usage), as well as CPU and network utilisation, and none of these show any signs of trouble when the issue arises. I would have thought our configuration isn't that uncommon, so I'm wondering if anyone else has seen this, and managed to resolve the problem?

    Read the article

< Previous Page | 193 194 195 196 197 198 199 200 201 202 203 204  | Next Page >