[NTLUG:Discuss] Why it can be argued MS is better than Linux (RANT!)

[David Stanaway]

One thing which is odd.  Sounds like the clock is cooked, or the battery 
on the mobo.

david at yezhov:~$ date -d "49710 days ago"
Fri Dec 13 14:45:52 CST 1901
david at yezhov:~$ date +%s -d "Fri Dec 13 14:45:52 CST 1901"
-2147483648
$ bc -l
2^31
2147483648

It is coming up with the largest negative 32 bit timestamp possible as 
the date of the last fsck.

If it keeps complaining on /home, but the rescue disk checks it fine, 
try tunefs (The appropriate one for the fs you are using, which would 
have been a very appropriate thing to mention!) and make it not check 
the filesystem after a certain amount of days, and check the options for 
the mountpoint entry in fstab

EG: on tune2fs there is a -T option you can use to set the time it was 
checked.


I am not sure what binary format the timestamp on the filesystem is that 
stores when it was last checked, might be something wrong there. Is the 
disk an IDE or SCSI disk, if IDE, it probably has S.M.A.R.T firmware, so 
check the smartmontools. Also check the date on the system when it boots 
up. If the battery is dead, and the clock initializes to 1901, then when 
you get online ntp fixes your clock up to current, that might have some 
issues. Operating linux with the clock really wrong can mess things up 
esp if you consider that unix time is seconds *since* 1970. Maybe there 
is a date bound bug in fsck (Unlikely, but maybe)

Stephen Davidson wrote:
> Greetings.
> 
> I never thought I would be writing an email with this subject, but after 
> events that started yesterday at about 1pm, it is now pretty easy.  
> Actually, the events started some time ago -- but they culminated in the 
> biggest disaster I have ever personally experienced in my history as 
> computer professional.
> 
> I have just finished repairing a machine (Linux) that had suffered a 
> hard drive failure several months back.  Not linux's fault, and that was 
> a fairly easy fix.  The reason that I only now just finished repairing 
> it is that I have been occupied with dealing with other Linux failures.  
> The most recent of which was a printer port failure -- that took 6 weeks 
> to find the workaround (Note: workaround -- not Fix) for.  And only for 
> my printserver.  My laptop still does not recognize its printer port 
> (both machines are OpenSuSE 10.0).
> 
> Yesterday afternoon, due to several security advisories, I decided to 
> apply the outstanding security patches for SuSE Pro. 9.2 to my server, 
> located at a Colocation site downtown.  Mistake.  BIG Mistake.  By 2:00 
> it was apparent that the system was not going to come back up w/o major 
> intervention.  So, I good bye to spousal unit, and head out, with master 
> disks in hand in case something was thoroughly corrupted and needed to 
> be reinstalled.  Get downtown and over to server.  It was trying to do a 
> fsck because none of the filesystems had been checked in the last 49710 
> days.  20.4% through the fsck of /home, and it hangs.  Everytime.  Ok, 
> that's what rescue disks are for, and I had brought mine.  Pop them in, 
> run them, and everything works out smooth.  Woohoo, I think.  I will be 
> home in time for dinner!  Reboot system, no checks in 49710 days, 
> restarting checks.  Ok, no biggie, I think.  They passed the rescue 
> disk.  Hung at 20.4%.  At least with MS, when one version of Scan disk 
> says your filesystem is fixed, the others will as well.  So, rescue disk 
> again.  Rescan.  Everything still ok.  Fine.  telinit 3.  No network, 
> and messages about how the system can't find half the files it needs.
> 
> Ok, I think, this OS is cooked.  So I run SuSE AutoRepair.  First it 
> does is scan the partitions.  They all pass.  Then it tries to verify 
> the RPM database.  That crashes.  Literally.  I get a red box saying an 
> Error has occured.  So I try to reinstall 9.2.  Does not work, same 
> issues.  Reformat /boot & /usr, and reinstall.  Still does not work, 
> same errors.  Go get spouse & OpenSuSE 10.0 disks.  Try installing 
> 10.0.  After all, other than the printer ports, and a other issues I 
> have mostly figured out how to deal with, its been running 3 machines 
> for several months now.  CD 1 installs no problem.  Reboots, and now 
> can't find the CD drive to read CD2.  I try a few things, and keep 
> running into that 49710 days thing with the partitions, again.  Ok, now 
> 7pm, and both spouse & I hungry, so we go for dinner and to think things 
> over.  Decide to resize partitions on the Data Hard disk, and put a copy 
> of /etc & /var there.  I am guessing there is some kind of funkiness 
> with the kernel modules, and the way to go is to reformat /boot, /, 
> /usr, and /var.  So, get back to the Colo facility, and start resizing 
> partitions.  At which point, spouse gets really freaked out "Linux is 
> way too complex for end users, if it is this hard to fix when something 
> breaks!".  Resize works.  Copy works.  Start Install.  Select partitions 
> to format.  CD 1 apparently installs w/o problem.  System reboots to 
> start newly installed kernel.  Unformatted data partions, not checked in 
> 49710 days.  No hangs, but check fails as well.  It is now 11pm.  I have 
> been at this for 8 solid hours (not counting break for dinner), over 
> half of which with non-techie spouse keeping me company.  It is obvious 
> I am not going to get this machine up tonight.  So I make the 
> appropriate arrangements, and pull server.
> 
> http://j2eeguys.com.  Its still offline.  There is a hole in the rack 
> where the server is supposed to be.  This email address is a redirect to 
> my j2eeguys.com server.  The server is on my desk.  Now that I have 
> enough diskspace online (as of this morning) to back it up again (btw, 
> backup server still full from all the crashes on other HW, I have not 
> had a chance to clean it out yet), I will backup said server -- User 
> space needs 30 GBs, which is more than was available on my laptops.  
> Once I format all the disks and partitions, maybe it will accept an OS 
> and boot again.  Then I wonder how long it will take me to reload and 
> reconfigure all the services that are needed.  I can't use the old 
> configurations -- they have all changed, and all attempts to upgrade 
> fail, as an attempt is made to write fstab during the upgrade, which 
> always fails.  For searching Google?  I have no idea how to describe 
> what is going on in less than one paragarph, let alone a single 
> sentence, which is what Google is looking for.  Mailing list?  Well, 
> that requires the ability to receive email, which I don't currently have.
> 
> Maybe I will get email back by the end of the week.  Who knows?  
> Meantime, my client has big design meeting today that I need to attend.  
> So, this whole thing needs to get shelved for this morning.
> 
> I have never had this much trouble from my MS machines.  If things get 
> so bad a reinstall is needed, 1.5 hours is the most I have ever had to 
> spend, plus about the same for the services -- not 8 hours, and still no 
> luck even getting the install to work.
> </end rant>
> 
> Regards,
> Steve
> 
> --
> Java/J2EE Developer (NOT a Linux Guru!)
> 
> 
> _______________________________________________
> http://ntlug.pmichaud.com/mailman/listinfo/discuss
>