Re: mysqld wont start...

Johan Lindquist wrote:
So anyway, it was like, 03:00 CEST Jul 31 2008, you know? Oh, and, yeah,
The Natural Philosopher was all like, "Dude,

dear boy, I have been running unix servers for years [..]

Good for you! And, may I say, it's a wonder this is the worst that has
ever happened to you in all those years given your flippant attitude
towards infrastructure.

Experience, rather than opinion, shows that what kills boxes is bad
disks, and/or bad memory. Not losing power.

The only times I have lost PC's or *nix machines have been when
power goes down *when booting them*. As happened in this case due to
a very unusual combination of events.

[..]

Apparently you realise that problems can, and do, occur when losing
power, yet you get all huffy and condescending when I point out that
an UPS would save you the embarassment next time the power goes out.

R, actually it was you who got patronising and condescending when I challenged your attitude that throwing mone at a problem wold make it go away.

"An engineer is someone who can do for sixpence what any damn fool can do for a quid"


All the UPSs in the world wont help if a machine crashes from a hardware fault.

You appear to have no idea about calculation of risk/benefit/cost.

In short you are a modern 'box ticker'


Just like the guys who ended up with egg all oer their faces with their 'diverse routed backbones via different carriers' when a digger ripped up a duct in the middle of England, and guess what, all the diverse routes and diverse suppliers of optical fibres were renting bandwidth off the same set of cables..


computers crash and get damaged data wise for many reasons other than power failures. A UPS is no good if someone trips over the power wire to the computer. Are your power wires laced with tie wraps to the chassis? Ours were after the first time. Cost nothing. And the cables were routed into cabinets and re-engineered not to be too long.

Computers crash because disks get old and wear..we had a policy of replacing disks after 3 years, on busy servers, irrespective of the actual condition..those disks were recycled int less demanding areas.

WE did teh calculations on waht we would have to spend to keep the whole machine room up and running for 12 hours..it simply wasn't feasible without massive amounts of money. Or a diesel generator. I have been to a site..on an island, where teh TElco does in fat have TWO sets of diesle generators, and the while basement of two buildings is a massive oil tank for them: THEY can get cut off for a week from the mainland..thats what it takes to keep a telehone exhnage (and indeed an island internet) going..


Backups are good: I hold my hand up to that. Backups mean a quicker recovery from ruined disks. But these days i run a free service on a tight budget.

But the old days of reusable tapes are gone. Data is too huge. It makes more sense to have a spare machine or disk.

It's sort of funny, but as I said, if you're that stubborn I guess
you'll just take longer to learn. Or not learn, as it were.

Long experience does not seem to help in making people open to good
suggestions in general, it seems you're no exception to that rule.


I think in fact, you are describing yourself: Your thinking is sloppy. I said 'power falure during boot corrupted my files' and you say 'get a UPS.'. Not only is that patently not answering the question I posed, it actually doesn't even really totally protect against the classes of occurrences that cause computers to lose power or crash during booting.

My google research into this reveled lots of people have had this corruption, mostly when they had to hard reboot a computer that had hung during boot, often when upgrading to new software.


An UPS won't protect from that.

Anyroads, it's none of my business, really, and I was just offering
some helpful advice based on /my/ humble experience, so best of luck!


humble? ha!
.

Relevant Pages

  • FS: Lots of Digital Alphaservers and related equipment
    ... Dual power supply ... x SAS FDDI ...
    (comp.sys.dec)
  • Re: A sorry saga
    ... May we assume you have the XFC cache going? ... Within hours, however, we were seeing disk and file corruption - I cleaned ... corruption when the power had gone. ... We then scrounged some disks & cradles for inside the ds20e, ...
    (comp.os.vms)
  • RE: Aborted commands with arcmsr and 2xWD1500ADFD in RAID1
    ... Aborted commands with arcmsr and 2xWD1500ADFD in RAID1 ... Putting the disks in anti-vibration mounts in 5.25" slots. ... Using legacy ATA power connectors instead of the SATA ones. ... > The vibration could be caused by FANs not only by themselves. ...
    (Linux-Kernel)
  • Re: Power saving/spinning down RAID drives
    ... If you don't use the server it will take up memory (I have turned off ... > get some power savings for nothing that would be nice. ... And then there are the disks. ... That entails turning off swap, natch, and going noatime, and so on. ...
    (comp.os.linux.hardware)
  • Re: Aborted commands with arcmsr and 2xWD1500ADFD in RAID1
    ... Turning on/off read-ahead cache in the Areca RAID. ... Putting the disks in anti-vibration mounts in 5.25" slots. ... Using legacy ATA power connectors instead of the SATA ones. ... >> Hi Aron, ...
    (Linux-Kernel)