ARCHIVE: June, 2011

Firefox Plumber Eliminates Memory Leaks

If you’re a power user like me, you’ll find Firefox will often balloon to about 1 gig of RAM usage after some time. Often times it’ll hit that point and then just crash.

I’m probably in the minority but my surfing habits tend to be open a mess of tabs and then come back and read later. I never go back to bookmarks (even though I do have a delicious account and actually tag stuff). At any time, I may 25-100 tabs open in my browser.

I’ve had to install add-ons like session manager just to ensure I don’t lose my tabs when working because Firefox’s built-in tools wouldn’t restore my tab session when it crashed.

I recently stumbled upon Firefox Plumber (http://www.rizone3.com/2011/firefox-plumber/). I’ve been testing it for a few days now and I have to say I’m completely blown away by how well the utility works.

firefox plumberCurrently I have 32 tabs open in Firefox. Under normal loads, the browser would utilize between 800MB-1GB of memory. With Firefox Plumber running, which is only utilizing about 708k in memory usage, Firefox is currently fluctuating between 4 and 7 MB of memory usage.

The only caveat is that Firefox Plumber does utilize about ~5% of the CPU to keep Firefox tamed.

After some sleuthing, I found it accomplishes this by offloading the memory to the swap file which could present it’s own issues. I’m going to stick with it for now since Firefox appears to be more stable and see how things work.

Detecting ASP.NET debug mode

The Problem

Recently I ran into a situation where I needed to debug ASP.NET code in a production environment that we had no control over. The server was managed by a third party support team and we deployed to a staging environment through a custom web deployment utility they built.

Of course, the code ran locally and on our internal staging environments with no issues but when deployed to the client’s remote staging servers, the application was encountering odd errors that we couldn’t replicate.

At this point, I wanted to add code to the web application that could be turned on and off without having to recompile deploy new dlls because of code changes in the code behind. With this particular client, code changes would trigger security scans that took over a week to complete and we were short on time.

The Solutions that Should’ve Worked but Didn’t.

Page Tracing wasn’t working. I remembered the #if Debug and HttpContext.Current.IsDebuggingEnabled statements worked rather well in other projects.

So I added:

to the web application. Nothing happened so I tried:

but it kept returning false even though debug mode was set to true in the web.config file.

The Solution (that worked!)

Finally I got the bright idea to read out the debug setting out of the web.config and execute code if the debug flag was set to true. How to do so wasn’t exactly obvious though.

After some searching, I finally figured it out and here’s the code snippet that will execute only if the debug flag is set to true:

Why Didn’t the Normal Debug Statements Work?

The issue was that the machine was configured as production. The machine.config overrode the web.config since the <deployment retail=”true”/> switch was set in Machine.config.

This setting disables Page.Tracing & #If Debug and always sets the IsDebuggingEnabled to false. This was done by Microsoft as a security precaution for companies so they could ensure no applications were deployed with debugging enabled by mistake.

Bonus! How Do I Loop Through All Session Variables in C#?

I wanted to see what the session variable values were during execution of the page with the caveat that it would only run if the debug flag was set to true.

I added the code directly to the bottom of the aspx page since I didn’t want to modify the code behind and voila! Once the code was added to the page, we found that the expected session variables weren’t populating correctly on the remote server. Unfortunately it required a code change to resolve the issue but I never would have found the cause without the above snippet of code.

Stupid Admin Tales Part 1

Life as an system/network admin can be extremely fun and satisfying when you’re not bogged down with management and people breathing down your neck. Of course it has moments to cause you to sweat a giant puddle in the middle of the server room. We’ve all made mistakes and (hopefully) learn never to repeat them. Sometimes we’re doomed to repeat them no matter what precautions we take.

In one of my first jobs as a system admin, I used to be responsible for a small business server in a 5-10 user office. One of the downsides of working in a small business is often the budgets don’t coincide with the real needs and you’re often forced to make things work using bubble gum and sticks. Duct tape was a luxury for spoiled admins that was completely out of the budget I was given. The first machine purchased was a bare-bones Windows 2000 machine which served as a file and print server for the office. Not too bad, right? Unfortunately due to budget constraints, this machine ran Windows 2000 Professional, not the server edition that was recommended. It had to function on a workgroup as a server since Active Directory was not an option. Security was managed at a workgroup level meaning all changes had to be made on every PC individually as well as the server. Luckily with 5-10 users, it wasn’t unmanageable and changes could be made to most machines after hours.

As the business grew due to better use of the technology and skills of the IT team (read: Me), the budgets increased slightly and I was allowed to upgrade hardware to a better machine but the Server license was still out of the budget I was provided. The network still purred and all users were happy with the performance and uptime and how smooth things ran. As more data was used and saved, backup became a major priority. With the limited budget a tape backup drive was too expensive, and as this was pre-cloud era, a Maxtor One Touch backup and DVD backups were the only solutions available as options to consider. Dual backup systems were a must for redundancy and off-site backup capability. Everything was implemented and tested successfully with restores working with no issues from both the drive and DVDs.

Flash forward roughly two years and the server’s primary hard drive fails and the secondary seemed to have become corrupted. Luckily the server was under warranty and the hard drive was replaced at no cost. There were backups of everything so data loss wasn’t a concern. After replacing both drives, I loaded the Windows disk and began the install process. Setup detected the new drive and my standard operating procedure is to format the drive to get it NTFS ready. The C: drive was selected and setup began the format and I walked away to complete other tasks. I came back a short while later and found Windows was installing and smiled. It was about then I noticed the lights on the Maxtor drive blinking as if data were being read/written.

A frown replaced the smile as my brain tried to process why the light would be blinking if Windows is installing on the drive and hadn’t gotten to the driver installation portion yet. I processed different scenarios as quickly as possible trying to find valid reasons why the lights would be blinking. It was a horrifying realization that there’s no way to cancel the install without shutting down the machine forcibly which could damage a drive. I weighed my options carefully and decided that in the event that my fear was for naught, I’d simply be able to start the install process over again.

Off the machine went and the Maxtor drive stuttered. Sweat began to build on my forehead as I knew there was no denying it. Windows setup was inexplicably installing to the external drive even though I selected the C: drive. I began damage assessment to see how bad things were. I unplugged the drive and reinstalled Windows and loaded the drive back on. All the data was gone and a partial Windows install was all that remained.

“Wait! Maybe data can be recovered using one of the many tools in my arsenal!” I so foolishly thought to myself. Windows had somehow managed to install itself over only the sectors where all the data was and only a few files were recoverable. I then realized I had DVD backups and quickly rushed to retrieve them from my office. I plopped the most recent disk in and then tried to copy the data back. A message box that simply said “Cyclic Redundancy Check” suddenly greeted me. I grabbed the next disk and tried to restore from that to find the files wouldn’t copy or open. I grabbed the first disk that I tested and knew worked only to find even the files there wouldn’t copy or open. I was dumbfounded as I had tested the discs to ensure that the backups were valid.

So at this point, you might be asking yourself what could possibly have happened? It turns out for some completely inexplicable and idiotic reason, Windows setup chooses the external drive as the primary and sets it to C. The DVD backup issues I only figured out recently. The issue was caused by the NTFS ID being different for the new Windows install. The NTFS IDs were now different on the new server. As the data was on non-writeable media, there was no way to set permissions of the files which made them completely useless.

Lesson learned? Unplug all drives when doing any OS work and DVD backups aren’t worth the disks they’re saved on.

Years later, a friend called me up with issues with his PC and asked if I could help. I went over, diagnosed that the hard drive was failing and that it needed to be replaced was done with no issues. After reconnecting all the cables back to the PC, I checked and saw no backup drives anywhere. I double checked and asked if said friend had backups of the data to be restored and was assured he did and that the drive was safe. I began the install and Windows began to format the new drive. It was then I heard the familiar grind of an external drive when data was being written to it. Reflexively, I shutdown the PC and cut off the installation. I called to my friend and asked why I heard an external drive when none were around that I could see even after tracing all cables. One of his many skills was carpentry and it turned out that he felt the drive was an eyesore and mounted it away behind the desk completely out of sight. I didn’t find any cables to it when I traced them all because the drive was plugged into a printer with a USB hub built into it. Even worse of a coincidence, the new drive wasn’t recognized by Windows due to incorrect jumper settings. The single drive I saw in the list which assured me there was only one drive available turned out to be the external drive.

I spent about two weeks recovering the data on that drive. Luckily I only lost some unimportant videos.

Lesson learned? Unplug all USB cables until after Windows setup is complete.

There are no more results.