I have been having some issues recently with my windows 7 system running extremely slow and becoming very unresponsive, everything would take ages, opening programs, searching, start menu, sometimes even clicking the mouse would have big delays.
Task manager showed I still had plenty of RAM and CPU though, having had such situations before I have learnt from experience  that the problem is usually caused by disk access, some program is thrashing your disk causing a bottleneck on the rest of your system.
On windows 7 this is pretty easy to spot as you can run the “Resource Monitor” and look at the DISK tab and see what programs are doing the most reading and writing, and often you will also see a bunch of hard faults being caused as well.

In this case I discovered that Google Chrome was the culprit, now the only thing that chrome really needs to access disk for is the cache, so that seemed like the obvious starting point, so I cleared out my browser cache, and voila, problem went away. I then Googled the topic (irony) and found manhy others with the same issue, so it seems that Chrome does have a habit of thrashing your disk due to the cache, perhaps it gets too big ?

Shortly after this I started having RAM issues, I have 8GB RAM in my system and Chrome started consuming all of it according to task manager. On opening it would consume 1GB right off the bat, and once I had all my usual tabs open (about 10), this would jump to 3GB+, which is mental.
I had recently moved my system from HDD on to an SSD, and that is pretty much when the problem had started, so I was sure this had to be the cause.
As part of this migration I had followed a number of best practices for SSD usage, one of which is to not use a SWAP file, especially when you have a lot of RAM as I do.
Again Googling the topic found lots of others with this issue of Chrome consuming lots of RAM, and the general consensus seems to be that this is normal, Chrome uses a lot of RAM because each tab runs as a separate process to avoid a crash in one tab affect your whole browser, plus each extension consumes RAM as well.
With this in mind, it seemed that my lack of a swap file was most likely the cause here, Chrome probably uses this much memory all the time, but would usually be using virtual memory and thus the swap file so you don;t notice. So I re-enabled the swap file and rebooted my system, and voila, my memory usage has reduced by about 2GB.
The downside of this is that SSD’s have a lower lifespan than HDD and this is the reason for not using a swap file, to reduce wear on the drive, but it seems this is not viable if you use Chrome.
The other option is to reduce the number of browser tabs you have open to save memory, or install more RAM, but honestly who would think 8GB is not enough for a browser ? I think Google seriously need to work on Chrome’s memory usage, I may well have to switch to another browser now.

Sociopaths are everywhere,  according to research at least 4% of the population are sociopaths, the most common types you are likely to have come into contact with or be aware of are "Rogue Traders, Conmen, Cowboy Builders etc",  these are the worst types who actively go out out of their way to intentionally harm other people with malice, but sociopaths also exist in our everyday life in less obvious guises, and unknown to you may even be one of your friends. 
I have come across a few in my life, probably more than I am aware of, but the most surprising of which turned out to be one of my best friends of more than 10 years, which just goes to show how clever they can be and how long they can deceive you for.

The main characteristic of a sociopath is a complete disregard for the rights and feelings of others and a complete lack of conscience when causing someone (even a friend or family member) physical or mental harm. 
Sociopaths are also unable to conform to what society defines as a normal personality (although I would argue as to what is normal these days). Antisocial tendencies can also play a big part of the sociopath’s personality. This pattern usually comes into evidence around the age of 15. If it is not treated, it can develop into adulthood.

Visible symptoms include physical/verbal aggression and the inability to hold down a steady job or relationship. The sociopath typically shows no regret or remorse for his or her actions and will never accept any responsibility or blame for his/her actions, often believing they have done nothing wrong and it is them who has somehow been wronged.

Although these symptoms may all present, they may not always be evident and may not always be malicious or extreme. Research has shown that the sociopath is usually a person with an abundance of charm, wit and often intelligence. He or she may come across as friendly and considerate, but these attributes are usually superficial. They are used as a way of blinding the other person to the personal agenda behind the sociopath’s behaviour.

Read more...

I recently upgraded a server to Railo 4 and noticed my catalina folder started to fill up with random folder names, e.g.

C:\railo\tomcat\conf\Catalina\acoaqnyfnl (or whatever the equivalent is on Linux)

Every time I checked back the number of randomly created folders had increased exponentially, I ended up with thousands of them before I found the cause. This can cause multiple issues if left to grow, affecting the performance of Railo and Tomcat and your server in general once you get seriously large numbers of files or folders.

I poste don twitter a few times but unfortunately not one single person in the Railo community had a clue on this one so I was rather stumped initially.
Then it dawned on me, this folder normally only contains an entry/folder for each of your virtual hosts, so I went and checked another server and noticed that not only did it contain a folder for each virtual host, but also for any other domain alias that pointed at those virtual hosts, even if they were not defined in the server.xml.

Once I realised this then the reason why the folders were being created became obvious, any host header that is used to access any Railo site on your server will cause a new folder to be created in c:\railo\tomcat\conf\Catalina, at this point I don't actually know why this happens only that it does. Previously I thought that any aliases had to be defined in the virtual host config, but this is obviously not the case.

So the next question is why are all those host headers getting through?

This one was simple too, the default website on IIS is set to respond on <ALL Undefined> by default, meaning that it doesn't expect any host header and will display the default website for anything that points to any IP on the server if there is no other site with a valid entry for that host header. And of course Railo is installed to the default website as the default webapp.

So the next question was, why all the random folder names, which clearly were not normal host headers as they were not domain names. My first thought was that it could be bots trying to hack the server using randomly generated host names, so I checked my IIS logs and found a bunch of entries like this.

2013-02-06 07:08:10 W3SVC1 BTI-APP1 1.2.3.4 HEAD / - 80 – 5.6.7.8 HTTP/1.1 Mozilla/5.0+(Macintosh;+Intel+Mac+OS+X+10_8_2)+AppleWebKit/537.17+(KHTML,+like+Gecko)+Chrome/24.0.1312.57+Safari/537.17 - - kcqtvklylt 200 0 0 259 249 3430

As you can see the request was coming form a MAC and was using the host header "kcqtvklylt", which explains the folder names.

Whether this was dodgy or not I really could not tell, the logs did not show any specific filenames being requested or any query strings such as sql injection, but I guess it certainly could be a trojan or malware of some sort scanning servers for open ports, which seems more likely than a hacker using a Mac

The solution was simply to put a host header on the default website to stop these requests even getting through the web server, once I did that the problem went away and no more randomly created folders.

and the moral of this tale, if you are running Railo or anything else on Tomcat, always use host headers (bindings) for all your sites, do not have any setup to allow <all undefined> or "*" without any host header at all.

One of the things that really grinds my gears is lousy customer support, and sadly pitiful customer support seems to be the norm these days, they simply don't care about their customers or providing a good service, so I have decided I'm going to start blogging my experiences and review the best and worst companies I deal with. Below are my experiences for last year.

In this article:-

• DELL
• SwordSky
• WHMCS
• Gradwell.com
• SKY
• KAYAKO
• Zen Internet
• O2
• ENOM

For the record I know full well that it is not possible to please all of the people all of the time, and legitimate mistakes and misunderstandings do happen, so I try not to judge any company based on just 1 poor experience unless it is so bad I simply do not want to do business with them ever again, all of the companies mentioned below are ones I deal with on a regular basis. 
Now it has been suggested that perhaps my standards are are too high, but I disagree, my requirements are very simple and I  have consistently managed to provide this high level of support to my own customers for the last 12+ years, even from my days as a sole trader when it was me doing everything by myself, and during those 12+ years I have also consistently received good reviews and feedback from customers, in fact I can count the number of unhappy customers we have had in all those years on both hands.

Here are my (simple) requirements:-

• respond to emails/tickets in a timely fashion (my company replies to most tickets within 4 hours)
• when dealing with email, actually read it properly and provide a response that is actually related to the question asked
• if you don't know the answer, ask someone else and try to find out, don't simply ignore the customer/ticket and hope they will go away
• actually give a damn about what you are doing

I have generally found the worst performers is where support is being outsourced to India, and the reason for this is in most cases is due to a poor understanding of English (odd considering most of their customers are English speaking) and as a result they do not understand the questions they are being asked or simply do not read it properly, and as a result give responses that are total dribble, unhelpful, often using incomprehensible broken English and unrelated to the question the customer asked. 
For small companies it is understandable why they outsource as it is the only way they can afford to offer 24/7 support, but for big companies like Dell and O2, I do wonder why they do it, as it only serves to harm their reputation. 
Many of these common issues could be handled by some simple quality control, monitoring of support staff and their responses and pointing out mistakes so the staff learn by them and improve their communication skills and responses. And yes this is also something I do at my company and it has worked well. 
However sadly this is something most companies clearly do not do, especially the big boys, leaving the issues to continue and customers to become more frustrated.

That said, some of the worst support I have experienced is "in-house" provided by English people and and not outsourced at all.

Read more...

Had an issue this week where all SOLR collections on one of our ColdFusion 9 servers stopped working, and SOLR was not accessible from the CFADMIN either, with none of the collections showing.

Trying to add a new collection in the CFADMIN would result in this error

An error occurred while creating the collection: org.apache.solr.common.SolrException. Check the Solr logs for more detail.

The websites would be giving the error

The collection COLLECTION_NAME does not exist

Having never used SOLR, this took a fair bit of digging to resolve, Google turned up plenty of other people having this error but no solutions.

It turns out that whenever a SOLR collection is created, a set of configuration files are also copied into the location of the collection.

E.G.
You create your collection at
D:\wwwroot\mydomain.com\wwwroot\SOLR\collections\mycollection

Inside this folder will created a "conf" folder which contains a bunch of config files which SOLR needs to read in order to initialise and use your collection. these are copied from C:\ColdFusion9\solr\multicore\core0 on a default installation.

The first file it looks for is solrconfig.xml, if it cannot find this file then it will throw a configuration error and the default behaviour of SOLR is to simply abort when this happens, thus SOLR completely stops working.

To find out which collection is causing the abort you can check the SOLR admin, which by default is on http://127.0.0.1:8983/solr/ (thanks Jeff).
This will give you an error showing you which collection is causing your problem, the error should look like this

java.lang.RuntimeException: Can't find resource 'solrconfig.xml' in classpath or 'D:\wwwroot\mydomain.com\wwwroot\search\collections\mycollection\conf/', cwd=C:\ColdFusion9\solr

There are 2 ways to fix this.

1. Restore the missing files to the above folder from your backups.
2. Remove the collection form SOLR
   open C:\ColdFusion9\solr\multicore\solr.xml
   And remove the offending collection.
   Restart the SOLR service.

Now you should also notice on the SOLR admin error that it says:-

If you want solr to continue after configuration errors, change:

<abortOnConfigurationError>false</abortOnConfigurationError>

in solr.xml

This is actually untrue, I tried this and could not get it to work, so went and posted on the SOLR forums only to discover that this feature was actually removed long ago, even more odd is that it was voted to be removed by most of the community. Why they would vote to remove a feature that would stop SOLR from crashing and make it more stable is beyond me.

So what this means is that SOLR is completely unsuitable for shared hosting, where other users are going to cause this problem all the time, if someone deleted their collection, deleted their website due to rebuild or moving hosts, any action that removes the collection files is going to break SOLR.
The only safe way to remove a collection is with <cfcollection action="delete">