This is what I do too. Homebrew hardware running Debian, with 4 drives in a RAID5 configuration running EXT3, shared over the gigabit network using Samba. Bottleneck is the laptop hard drive. It works great, and I can afford to have a single drive fail with no data loss. I mainly use it to back up hard drives and to store media files (movies and music.) The server also doubles as the network's router, but that's irrelevant to this discussion.
My servers and workstations increment through the alphabet.
Addled
Burnout
Clueless
Dimension
Eptitude...
Last octet of the static IP is based on the position in the alphabet of each hostname and the NIC number in the system (Addled =.10, Burnout =.20 and.21 etc) regardless of which subnet that NIC resides on. When a system is retired, so is the name. Using this method I can get a pretty good idea of how old a system is in relation to the others. This scheme only works in a very small environment.
My suggestion is thus:
1-25 workstations: be creative. Use names that you can remember.
25-100 workstations: create groups of rooms or departments based on themes, and be creative within the themes. Tag hostnames on cases.
100+ workstations: use an asset-tag type scheme, possibly including location (campus-level, not cube-level) information that won't change, and perhaps user. Tag hostnames on cases. Reimage and rename systems when repurposing. Use a database to track specifics of the systems.
Assuming the disk works at all: Work on a clone, not the original.
If you are working on a 2nd generation clone you can afford to take risks in restoring
the filesystem. "Oh it that didn't work, fire up another clone and try something else".
This is what I have done using Ghost. There was an option to take a bit-for-bit copy of the drive. It didn't always work, but it would sometimes let you make a backup of the bad drive and restore it onto a good one, and then take a proper ghost image if necessary and recover your files. Any files which were corrupted on the original would still be corrupted, but anything else on the drive was suddenly readable and writable, meaning you can not only recover your data but even repair Windows if necessary and keep using the same drive image so there was typically no need to completely rebuild the system.
This appears to be a Caterpillar 789.
If so, It has a 1800hp 69 litre 16 cylinder V-engine.
112 Tons empty, with a load capacity of 180 Tons
Looks like they sell for about 2.25 million USD.
"It lost $153 million during its 2008 fiscal year and its current bankruptcy filing lists $390 million in assets along with nearly $527 million in liabilities.
As part of its proposed purchase of SGI, Rackable has agreed to take on an undisclosed portion of those liabilities. But there's a chance Rackable may not wind up acquiring the company."
The name alone could be worth $25 million if managed properly, but that $527 mil debt really bumps the actual cost of the property significantly. Also with a cashflow that seriously negative, they'd better have some really quick plans to turn this operation around. I would expect to see a lot more marketing for SGI systems if this goes through, probably within months.
[SGI] lost $153 million during its 2008 fiscal year and its current bankruptcy filing lists $390 million in assets along with nearly $527 million in liabilities.
As part of its proposed purchase of SGI, Rackable has agreed to take on an undisclosed portion of those liabilities.
Most people in geeky professions love feeling smart and useful. Then, when a user comes along asking them to be useful and show their intelligence (i.e. help) the geek scorns their lack of knowledge instead of being happy to help. Yes it might be the hundredth time you've answered that question, but just look at it as "this is why I'm the expert" instead of "I'm an expert and you you're asking me THAT again?"
It's not that you're always fighting with users but they all have a similar lack of common sense when using a computer, i would never drive anywhere if everyone exibited the same lack of common sense on the road.
You haven't driven in St. Louis in the rain, have you? With the first sign of precipitation, drivers become as stupid as the stereotypical computer user. Scary.
It seems like no other specialists have that problem on such a routine basis. When someone's doctor says "you have X disease" they generally don't look at him and say "no I don't." When an electrician says that something needs to be rewired, they might get a second opinion but they don't usually argue with the guy. Same deal with mechanics. With almost any other specialist it's understood that if you come to them, it's because you recognize that they know a lot more about medicine, electricity, or auto repair than you do.
You don't know many doctors or mechanics on a personal basis, do you?
Yes the apps hadn't been used in 16 hours, but what was the system doing with all that memory for those 16 hours? More likely than not, letting it sit unused is what. Why would you not want to leave the data where it can be accessed quickly? (This is NOT a rhetorical question; if there is a good answer, I want to know it.)
Excuse me? Why is this being modded up? There is not a single correct assertion in the entire post.
Virtual MEMORY is not-memory that appears to be memory, i.e. a hard disk page file.
Virtual ADDRESSING is a memory address that appears to be what is used but is physically elsewhere in memory.
Also, do you know how a hard drive works? There is always seek time in one of three types, relating to where the information to be read is located in relation to the current read head:
1) Same cylinder, same sector, different platter. Very fast (for a disc) because the only change to be made is which head is reading, so no physical movement is necessary.
2) Same cylinder, different sector. Not quite as fast because you have to wait for the platters to spin to get the correct sector under the read heads. Note it doesn't matter which platter because while they are spinning, the active head has already changed.
3) Different cylinder. This requires movement of the read heads and is the slowest operation. Note it doesn't significantly matter what sector the data is located on if the heads have to move.
Note I am not 100% familiar with the new orthogonal recording techniques, but I suspect those are even faster than changing read heads. Regardless, sequential data must still be located in close proximity to benefit.
Regardless, seek latency on a disk is never zero, though if sequentially accessed data is stored physically near (same sector when possible, then same cylinder if not, then adjacent cylinder, then the closest available cylinder.) it is faster than if it is located randomly somewhere on the disk. Being defragmented necessarily means you adhere to these recommendations as much as possible as your data is contiguous. Being fragmented typically means your data is scattered across cylinders.
It is true the pageFILE will never fragment, but the data within the file should be kept defragmented as it relates to a given program; data that is likely to be read sequentially should be kept together.
Physical memory has no seek latency, so fragmentation there has no meaningful effect (unless you can save some cycles by skipping address table lookups to locate the next read/write location.)
There are X frontends, but for the console you'll probably want to try aptitude. No package rating, but it downloads and installs just fine when you're done selecting your packages. Naturally it handles dependencies too. Look it up.
Speaking of stacking commands, you can stack greps to find, for example: "type file | grep foo | grep bar" in order to find both foo and bar on the same line in any order. Is there a way to find foo OR bar on any line (besides running two separate grep commands and merging the results, which loses the order)
Another handy feature: "type file | grep foo -A10 | grep bar" will find any line with bar in it which within 10 lines after an instance of foo. -A and -B (after and before) have wonderful utility for searching for events occurring shortly before or after other events in logs.
Not quite the same, but in a similar vein, cat/dev/urandom >/dev/dsp
Sometimes a quick white-noise machine is relaxing. Heck, I used that command in combination with 'at' to act as a makeshift alarm clock when I was just moving into my first apartment and had forgotten my only other electronic device with an alarm (my cell phone) at the office.
Now if only I could do that to a few servers in the data center downstairs. Imagine: Amidst all the fan noise, you're walking down the aisle and you hear... static. How do you troubleshoot THAT?!
CRTs contain a relatively large amount of heavy metals (mercury, lead, etc) and are rather expensive to dispose of properly (read: legally) so yes, many recycling places charge $50 to recycle a CRT. Your best bet is donating it to goodwill and writing a $20 donation off on your taxes.
Slashdot Mirror
This is what I do too. Homebrew hardware running Debian, with 4 drives in a RAID5 configuration running EXT3, shared over the gigabit network using Samba. Bottleneck is the laptop hard drive. It works great, and I can afford to have a single drive fail with no data loss. I mainly use it to back up hard drives and to store media files (movies and music.) The server also doubles as the network's router, but that's irrelevant to this discussion.
Sounds like a valid durability test for a SD card. I'd be willing to bet that it would survive longer in the pan than the bird would.
Audit trails are automatically generated, and can often be found on logs and cable trays.
My servers and workstations increment through the alphabet. .10, Burnout = .20 and .21 etc) regardless of which subnet that NIC resides on. When a system is retired, so is the name. Using this method I can get a pretty good idea of how old a system is in relation to the others. This scheme only works in a very small environment.
Addled
Burnout
Clueless
Dimension
Eptitude...
Last octet of the static IP is based on the position in the alphabet of each hostname and the NIC number in the system (Addled =
My suggestion is thus:
1-25 workstations: be creative. Use names that you can remember.
25-100 workstations: create groups of rooms or departments based on themes, and be creative within the themes. Tag hostnames on cases.
100+ workstations: use an asset-tag type scheme, possibly including location (campus-level, not cube-level) information that won't change, and perhaps user. Tag hostnames on cases. Reimage and rename systems when repurposing. Use a database to track specifics of the systems.
Assuming the disk works at all: Work on a clone, not the original.
If you are working on a 2nd generation clone you can afford to take risks in restoring the filesystem. "Oh it that didn't work, fire up another clone and try something else".
This is what I have done using Ghost. There was an option to take a bit-for-bit copy of the drive. It didn't always work, but it would sometimes let you make a backup of the bad drive and restore it onto a good one, and then take a proper ghost image if necessary and recover your files. Any files which were corrupted on the original would still be corrupted, but anything else on the drive was suddenly readable and writable, meaning you can not only recover your data but even repair Windows if necessary and keep using the same drive image so there was typically no need to completely rebuild the system.
Tacos schmacos. Try the water.
This appears to be a Caterpillar 789.
If so, It has a 1800hp 69 litre 16 cylinder V-engine.
112 Tons empty, with a load capacity of 180 Tons
Looks like they sell for about 2.25 million USD.
As part of its proposed purchase of SGI, Rackable has agreed to take on an undisclosed portion of those liabilities. But there's a chance Rackable may not wind up acquiring the company."
The name alone could be worth $25 million if managed properly, but that $527 mil debt really bumps the actual cost of the property significantly. Also with a cashflow that seriously negative, they'd better have some really quick plans to turn this operation around. I would expect to see a lot more marketing for SGI systems if this goes through, probably within months.
http://www.mercurynews.com/breakingnews/ci_12049610
[SGI] lost $153 million during its 2008 fiscal year and its current bankruptcy filing lists $390 million in assets along with nearly $527 million in liabilities.
As part of its proposed purchase of SGI, Rackable has agreed to take on an undisclosed portion of those liabilities.
...Suddenly not peanuts anymore.
I'm seeing "chick magnet"
...although you COULD go for the cute girl version too. I think someone mentioned mudflaps with the nude silhouette.
picture of a chick (baby duck, not cute girl)
+ picture of a stereotypical horseshoe magnet
Stories like this are why I don't check the firehose.
Why can't we mod entire "stories" -1 offtopic? This is quite possibly the lamest thing I have seen on Slashdot in years.
Most people in geeky professions love feeling smart and useful. Then, when a user comes along asking them to be useful and show their intelligence (i.e. help) the geek scorns their lack of knowledge instead of being happy to help. Yes it might be the hundredth time you've answered that question, but just look at it as "this is why I'm the expert" instead of "I'm an expert and you you're asking me THAT again?"
It's not that you're always fighting with users but they all have a similar lack of common sense when using a computer, i would never drive anywhere if everyone exibited the same lack of common sense on the road.
You haven't driven in St. Louis in the rain, have you? With the first sign of precipitation, drivers become as stupid as the stereotypical computer user. Scary.
It seems like no other specialists have that problem on such a routine basis. When someone's doctor says "you have X disease" they generally don't look at him and say "no I don't." When an electrician says that something needs to be rewired, they might get a second opinion but they don't usually argue with the guy. Same deal with mechanics. With almost any other specialist it's understood that if you come to them, it's because you recognize that they know a lot more about medicine, electricity, or auto repair than you do.
You don't know many doctors or mechanics on a personal basis, do you?
Yes the apps hadn't been used in 16 hours, but what was the system doing with all that memory for those 16 hours? More likely than not, letting it sit unused is what. Why would you not want to leave the data where it can be accessed quickly? (This is NOT a rhetorical question; if there is a good answer, I want to know it.)
Virtual MEMORY is not-memory that appears to be memory, i.e. a hard disk page file.
Virtual ADDRESSING is a memory address that appears to be what is used but is physically elsewhere in memory.
Also, do you know how a hard drive works? There is always seek time in one of three types, relating to where the information to be read is located in relation to the current read head:
1) Same cylinder, same sector, different platter. Very fast (for a disc) because the only change to be made is which head is reading, so no physical movement is necessary.
2) Same cylinder, different sector. Not quite as fast because you have to wait for the platters to spin to get the correct sector under the read heads. Note it doesn't matter which platter because while they are spinning, the active head has already changed.
3) Different cylinder. This requires movement of the read heads and is the slowest operation. Note it doesn't significantly matter what sector the data is located on if the heads have to move.
Note I am not 100% familiar with the new orthogonal recording techniques, but I suspect those are even faster than changing read heads. Regardless, sequential data must still be located in close proximity to benefit.
Regardless, seek latency on a disk is never zero, though if sequentially accessed data is stored physically near (same sector when possible, then same cylinder if not, then adjacent cylinder, then the closest available cylinder.) it is faster than if it is located randomly somewhere on the disk. Being defragmented necessarily means you adhere to these recommendations as much as possible as your data is contiguous. Being fragmented typically means your data is scattered across cylinders.
It is true the pageFILE will never fragment, but the data within the file should be kept defragmented as it relates to a given program; data that is likely to be read sequentially should be kept together.
Physical memory has no seek latency, so fragmentation there has no meaningful effect (unless you can save some cycles by skipping address table lookups to locate the next read/write location.)
$>test
Creates an empty file with name 'test'. Quite handy sometimes.
so does "touch test"
There are X frontends, but for the console you'll probably want to try aptitude. No package rating, but it downloads and installs just fine when you're done selecting your packages. Naturally it handles dependencies too. Look it up.
This is correct. For example: bash supports it, sh does not.
Another handy feature: "type file | grep foo -A10 | grep bar" will find any line with bar in it which within 10 lines after an instance of foo. -A and -B (after and before) have wonderful utility for searching for events occurring shortly before or after other events in logs.
Not quite the same, but in a similar vein, cat /dev/urandom > /dev/dsp
Sometimes a quick white-noise machine is relaxing. Heck, I used that command in combination with 'at' to act as a makeshift alarm clock when I was just moving into my first apartment and had forgotten my only other electronic device with an alarm (my cell phone) at the office.
Now if only I could do that to a few servers in the data center downstairs. Imagine: Amidst all the fan noise, you're walking down the aisle and you hear... static. How do you troubleshoot THAT?!
Read Mail -Really Fast /
I'm a PC, and a human being. Not a human doing. Not a human thinking. A human being.
CRTs contain a relatively large amount of heavy metals (mercury, lead, etc) and are rather expensive to dispose of properly (read: legally) so yes, many recycling places charge $50 to recycle a CRT. Your best bet is donating it to goodwill and writing a $20 donation off on your taxes.