I have a customer who was sold a new server because his prior computer service told him “his server was full and things will start getting randomly deleted to make space”
This service is now gone because of some unrelated issues but he was asking me and I honestly had no clue, servers are usually so overbuilt that such things have never been an issue in my world.
Would there be some kind of error message saying the users storage space or mail space is full?
Does some kind of first in first out delete process start taking place or does mail just start bouncing back? Files refuse to save?
If possible point me at a tech net or KB article describing how a win2k3 server manages such a situation so I can produce some kind of documentation for my customer to be able to demonstrate his last PC guy is full of …useless random data…
Well, I had wrote an entire post, then noticed you are talking about Windows 2003. In general, the guy who sold your customer on the fact that they needed a new server because it was getting full is either a great salesman or pretty much clueless. That said, if you run out of space on the primary partition (where the OS and cache are located), bad things could certainly happen, up to and including the server locking up or even blue screening. Assuming whoever built the original server knew what they were doing and the problem is that the users data partitions or the Exchange email partition is running out of space, the users will get error messages and the mail store might lock up or refuse to mount until you clean out some of the junk. It won’t just start deleting random files though.
This isn’t to say that having the server run out of disk space is a good thing, mind.
Depends on the OS, but generally if the users storage is full (assuming it’s not in the same place where the servers OS and cache are), the user will get an error message when they try and write a file to the share (or whatever) on the server, telling them that there was a write error, or that the drive was full.
You can set up some mail servers to cull old emails, depending on what you are running, but it won’t be random. Some email servers will crash if the server runs completely out of disk space, depending on how the OS and email software were loaded, but generally you’ll just get an error if they run out of space if you set it up right. And with some systems you can set thresholds to alert you that they are getting close, and also some that will delete old messages.
I don’t have a tech net article handy on this, and it’s been a while since I’ve done a lot of systems work, but if not one else gets back to you I’ll see what I can dig up on documented effects of running a Win2003 server (and presumably 2003 Exchange) out of drive space. You will probably be on fairly solid grounds though for pointing out to the customer that this other IT guy pretty much either lied to them or doesn’t know what s/he is talking about.
It’s one thing to upgrade a customers server because the hardware is getting older, or it just doesn’t have the drive space needed, or because the customer needs more hardware (or an upgraded OS) for other, requirements driven reasons…but to tell a customer that they need a new server because they are out of drive space and that files are going to start randomly being deleted? :dubious:
your post is petty much what I expected, the last thing I would expect is a serverOS to start willy nilly tossing files or emails to the great beyond. I was hoping for kb support because the last Pc place was a larger generally respected outfit and I would prefer to avoid ending up in the middle of any difference of opinion. A kb article or multiple online sources would give my “opinion” a lot more armor.
Even a raid array should be able to be imaged/cloned onto larger drives.
Are you talking about a disk drive on a server, or specifically Exchange?
I’ve never heard of win2k3 randomly deleting things. I can’t imagine where it would start.
Some of the apps we run on our servers have some settings we have to make sure are checked so that certain files don’t build up and use space. ASPEmail (a mail sending object, not a mail server app) has a “failed” folder we have to make sure stays clean. Some other things we’ve had to turn off specific logging so that the logs don’t get out of hand. You just turn on the logs when troubleshooting.
Other stuff we’ve had to write scripts for to keep cleaned out. I’m not sure if we had to write our own SQL backup delete scripts or if that’s built into the SQL backup service. I know the biggest thing for us is to have a script that cleans out all of the IIS log files after 10 days or so. But most of our servers are Web servers.
We also wrote a script that sends out an email every time one of the drives on the server gets to be X% full. Very handy.
Frankly, I just go on the server and run WinDirStat on each drive to see what’s so full. Usually it’s blatantly obvious (like a log file directory that’s 20gb) and I just clean up as needed, and then write auto-delete scripts as needed.
In my experience, nothing really specific happens when your server runs out of space. Shit just starts failing, and then you RDP into it and see “oh shit, the drive is full!”
Now, if you are talking about what happens when a machine dedicated to running Exchange gets full…that’s a different question.
Whatever the case, you’re going to want to figure out WHY the server is full, deduce whether or not it will keep getting itself fulled up, and figure out how to keep it from happening again.
This is bollocks. Windows Server doesn’t like it at all if a drive gets full, and it can lead to heavy fragmentation on that drive. You’ll get errors from clients and applications and the server may go very slow but the server itself should stay up so you can go to the console and clean up the box.
Well, the first thing to note is that Windows is pretty good at generating Event Log events warning you that a file system is getting full before actually hitting the wall (at which point, the event log cannot be written, so you no longer see the errors :smack:).
And the second is that Windows does not delete anything when a file system is full (apart from emptying the trash folder) - things start to fail, but random deletion does not occur. However, from the users perspective, things may appear to fail randomly due to variations in how processes handle data and failures.
For example, Exchange or MS-SQL which pre-reserves data store space on disk may work on with a full volume until the last extent is filled (transaction logs may fail, though, if they are not circular). Applications that overwrite existing files on edit rather than using a write-rename-delete process may also work (at least, until the file grows past the original size). Applications that use logging also tend to fail early. Roaming Windows profiles are very sensitive to disk full problems, and can cause much heartache and frustration for an administrator.
Finally, network clients that receive a failure from a server can be confused by the error message, and may display an incorrect response - this can confuse the user, and may lead to data loss due to an incorrect action.
Files may not get deleted per se, but information could certainly get lost. For example, any given program may have an .ini file that cannot be written/appended to because there are no free clusters. At that point the program will lose its settings, especially the next time you run it.