I'm looking for some general advice how I should treat my system.

I think it can be said generally, that the internal components of every maschine are at their most vunerable when they are switched on due to the high electrical surge running through them. This is why for example neon light bulbs usually finally fail when you try to turn them on. I suppose the same is (still) true for hard drives. These are my main concern right now, because thanks to Linux my computer has been running for almost two months constantly. My previous Windows installation was giving me BSODs galore as I watched in horror how my first hard drive was consumed by it.

Still I am beginning to wonder when I should actually shut it down for maintenance, since the CPU, GPU and PSU fans get clogged up with dust etc. I'd also like to know what the average failure in time rates can be for the above mentioned parts, since a five year old PSU of mine caused some nasty collateral damage (another hard drive) when it recently failed entirely.

I'm also curious about my TFT monitor: I want to save energy (and prevent ghost images) by switching it off every time I leave my desk. But doesn't this also damage the TFT like too much rebooting kills hard drives?

I know, this isn't a typical hardware 'problem', but since the people here are technically more adept and somewhat concerned (dual booting, long uptimes, ...) so I thought I'd put it here anyway.

Well, I have 8 PC's running at this moment. They run 24/7, most run 100% cpu doing F@H in their spare time... So they use a bit of power. I have crt's and lcd set on a 10 minute shutdown... I worry about the lcd's especially Have to have them sleep when not in use due to power use... Of course 2 of the boxes don't have monitors, like the print server and router box...

I shut down to blow them out when I think about it, which isn't enough in this very dusty rural area... I think the print server went 4 years at one stage. You couldn't see the motherbord components (really old XT flip top desktop case)... Poor ol p200... This is a machine that has been running 24/7 for about 10 years now, tho it did start with a p120 cpu (which was upgraded, not a failure). Still has an old quantum bigfoot 5 inch 2 gig HDD... I blow it out every couple of years if I remember.

I have not had too much trouble with failures, but it happens. I get a PSU now and then, but then some have been going for years... The print server still has an old XT style PSU... Wow that thing must be old. I've lost a couple of AT PSU's and one ATX PSU this year on a Sempron machine. No damage from any of these. So the PSU's seem to last okay, but some less than 5 years and some go longer...

Most of the failures here are fans. PSU fans and video card fans - which sometime take out video chips. So fans are something I have to check now and then - which I always forget to do.

The last disk damage I got was when the primary IDE went on my p120 in 1999. That board is still going on the secondary IDE in the print server box. Had a hdd die in one machine recently, but don't lose a lot of drives...

No real numbers. Some things last forever. Others die a lot quicker than you want. Monitors especially. My 19" crt lasted a little past the 1 year warrenty period

Now you got me thinking. I don't think I have even looked at the latest router box which I set up about 3 years ago... So I think it might need a blow out...

Only moving parts in a PC are the fans and disks, and not surprisingly those are the components that I seem to have problems with the most frequently. Some of the cr@ppy fans I've been getting lately should have MTBF's measured in seconds not hours. Fortunately hard disks have improved greatly since the days of having to manually park heads on every shutdown; they don't cause me nearly as much grief nowadays.

In any case, guess your "blowing out the dust bunnies" maintenance plan depends on many factors: environmental conditions, availability of a failover or hot spare server to minimize the amount of downtime (a high availability cluster would be even better), management's attitude (if it ain't broke don't touch it until it catches fire kind of attitude), etc...

Hi,

It's your electric bill! Thermal shock is the worst enemy.

Before I retired, we had lab machines that would run 24/7. Occasional HD failures, psu failure but few MB failures over a 14 year period. Most of our failed single unit systems were due to individuals that power cycled the systems. I mean users that would power up the system then power down to then power up again in just a few hours. These I would tell the person(s) involved to just leave the machines powered until the end of the day. Prefer if they would just leave them on but if they must power down to do so on a wide cycle period. If they were going to be gone for an extended period then power down if they must. Most would just power down the monitor.

Once others found that this was a success then they too would cycle in this manner.

Most modern systems have standby or shutdown cycling available. So to manually control is not a problem. As for newer monitors or LCD displays, I would cycle these down either via the OS or just manually shutdown.

HTH!