Raid setup

[Mike Thompson]

I inherited a small network with a server that is obviously setup with raid. There are two raptors for the OS and 2 WD drives for data. Since it is ~5 years old, I suspect we're living on borrowed time, and if this server goes down, it will cost the office a lot of dough. How can I determine exactly how its setup now, and how to deal with a disk failure in the future???? Is SiSoft Sandra going to be a good bet to give me specifics, or is there another program better for analysis? Any help would be greatly appreciated. Mike

 

[pitzel]

What operating system? If its MS-Windows, just look in Device Manager. If its some form of Unix, then you'd want to use something like smartctl to read out the SMART parameters of the drives. Personally, I'd suggest that you find a time to shut the machine down, give it a good blast of compressed air spray to clean the dust out of it, and then do an inventory of what's there. If, indeed, "it will cost the office a lot of dough", then my advice would be, at this point, to bite the bullet, and have a professional come in and advise you on what it will take to be reliable and maintainable. Since you mention that the machine is set up with RAID-1 (mirroring), then that will protect you against mechanical failure of the hard drives, and well-constructed and maintained computers easily last 10+ years (except the power supplies -- the cheap lowest-bidder junk tends to fail early). Off-site backups are still a necessity. I would advise against installing packages such as this "SiSoft Sandra" unless you are knowledgeable and experienced with such. Good luck! In the future, I would suggest some form of redundancy, and real-time backup or synchronization for the entire server. For a small business, this can be as simple as setting up a script to run rsync over a DSL line to sync the main business server, with a computer running in the owner's basement.

 

[OVERKILL]

Grab the models of the drives through device manager and find a set of spares for it. Maybe dust it out with compressed air as pitzel suggested, and then just LEAVE IT ALONE. If a drive fails, replace it. That's the purpose of a mirror. Other than that, make sure you have a decent backup routine.

 

[rationull]

 Originally Posted By: pitzel

Since you mention that the machine is set up with RAID-1 (mirroring), then that will protect you against mechanical failure of the hard drives

Not to put too fine a point on it, but a mirrored RAID setup with 2 disks will protect you against mechanical failure of *one* drive.

 

[beanoil]

3 words,.. Backup, backup, and CYA. As stated, mirroring will protect you from a failure of one drive. A dedicated backup routine saves you from a catastophic failure of anything. Data sensitive or mission critical servers are backed up at least daily, weekly may work fine for you. You dont say how large the raid is, but drives are darn cheap, and you can never have too many copies of critical data. There are subscription data houses that will handle your data backups also. If a failure would "cost your office a lot of dough", you should weigh what a regular backup routine could save you vs. the lost data, and go from there.

 

[pitzel]

 Originally Posted By: rationull

 Originally Posted By: pitzel

Since you mention that the machine is set up with RAID-1 (mirroring), then that will protect you against mechanical failure of the hard drives

Not to put too fine a point on it, but a mirrored RAID setup with 2 disks will protect you against mechanical failure of *one* drive.

Yeah, no kidding, and its very unlikely a computer is just going to, in the middle of the day, start screaming, "I lost a hard drive, I lost a hard drive!". A sysadmin (or someone with at least some basic expertise) needs to be actively monitoring the state of the server, reviewing event logs from time to time, etc. A RAID-1 (or even RAID-5) system will just soldier on, doing its work, not noticeably degraded in speed -- without 'telling' anyone, unless software has been installed and configured to provide for email notification. Of course, any level of RAID does not protect against: a) Viruses; b) Malicious users; c) Spyware; d) Server admin messups; e) Power surges or controller failure; f) Operating system software bugs that can create filesystem corruption. g) Drives being filled up, which can/often causes data loss when someone tries to save something to a directory that's full. Seriously, if you're not a computer expert, this is not a place to DIY. A 5-year-old server, coupled with the modern tendency for people to use larger files, emails, etc., etc., may very well have some capacity issues.

 

[Volvohead]

 Originally Posted By: pitzel

Yeah, no kidding, and its very unlikely a computer is just going to, in the middle of the day, start screaming, "I lost a hard drive, I lost a hard drive!".

Actually, all the better RAID controllers have communication utilities that DO issue alerts when a drive drops out. You can configure E-mail alerts to the administrator (and others) for a broad variety of events. Even if you don't configure the communication utility, many will also beep like mad following a HDD failure if you're anywhere near the device.

 

[pitzel]

Well, of course, you can configure SNMP traps, audible alarms, email or pager alerts, etc., on many systems. But until they've actually configured, and configured to be directed towards someone who is in a position to help, they're useless. You don't need a full-time computer geek, but at least make a relationship with someone with knowledge/experience who can keep stuff going and plan for the future accordingly.

 

[PandaBear]

Given the age you should have a pair of spare around to swap out the bad drives when one goes down, then replace the working old drive with a new one since both of them probably wear together and would die soon after the other. Or you can do preventive maintenance and replace the aging drive with new ones, but do it one drive at a time to catch infant mortality. Mirror should buy you enough time to limp along and get your replacement up and running, but make sure you don't let it run with only 1 drive for too long. I second that virus and user errors would cause more headache than a hard drive failure.

 

[OVERKILL]

 Originally Posted By: pitzel

 Originally Posted By: rationull

 Originally Posted By: pitzel

Since you mention that the machine is set up with RAID-1 (mirroring), then that will protect you against mechanical failure of the hard drives

Not to put too fine a point on it, but a mirrored RAID setup with 2 disks will protect you against mechanical failure of *one* drive.

Yeah, no kidding, and its very unlikely a computer is just going to, in the middle of the day, start screaming, "I lost a hard drive, I lost a hard drive!". A sysadmin (or someone with at least some basic expertise) needs to be actively monitoring the state of the server, reviewing event logs from time to time, etc. A RAID-1 (or even RAID-5) system will just soldier on, doing its work, not noticeably degraded in speed -- without 'telling' anyone, unless software has been installed and configured to provide for email notification. Of course, any level of RAID does not protect against: a) Viruses; b) Malicious users; c) Spyware; d) Server admin messups; e) Power surges or controller failure; f) Operating system software bugs that can create filesystem corruption. g) Drives being filled up, which can/often causes data loss when someone tries to save something to a directory that's full. Seriously, if you're not a computer expert, this is not a place to DIY. A 5-year-old server, coupled with the modern tendency for people to use larger files, emails, etc., etc., may very well have some capacity issues.

A properly setup server will notify a responsible party in the event that anything serious goes wrong. I consider this part of the initial configuration. This goes hand in hand with drive redundancy, back-ups, power conditioning/UPS, surge suppression on WAN lines...etc. Of course when dealing with somebody's "on the cheap" office server that they've setup themselves... well then all bets are off in regards to its current condition and configuration as well as the degree of implemented protection.

 

[javacontour]

RAID protects you against hardware failure only. You still need a back up solution to help protect against human error deleting critical data.