Trials and tribulations of a new Admin.

nutcase · 7 Aug 2016 at 16:28

So the dying G5 won't give in. All the services are (I think) transferred off, and to test I shut it down - give it a couple of weeks and if no-one whinges about anything, wipe it (I have a backup of it anyway to retrieve any files hiding there but nowhere else).

So I turned it off. Half an hour it was back on. So I turned it off again. 2 hours later is was back on. Looking at the iLO log it always comes back on exactly on the hour (but not every hour as there was a 2 hour gap for one of them).

The power config is set to not auto power on, the NICs are set to not WOL.

Any idea where else to look?! OS is 2008 R2.

Deleted member 138126 · 8 Aug 2016 at 13:23

nutcase_1uk said:
Interesting - HP MSL drivers install nice and easy on Hyper-V server 2012 R2 without issue. When I've installed the Dell equivalent I had to amend the registry to pretend it was regular server 2012 R2 first.

You are installing HPE drivers on a Dell server?

Deleted member 138126 · 8 Aug 2016 at 13:24

nutcase_1uk said:
So the dying G5 won't give in. All the services are (I think) transferred off, and to test I shut it down - give it a couple of weeks and if no-one whinges about anything, wipe it (I have a backup of it anyway to retrieve any files hiding there but nowhere else).

So I turned it off. Half an hour it was back on. So I turned it off again. 2 hours later is was back on. Looking at the iLO log it always comes back on exactly on the hour (but not every hour as there was a 2 hour gap for one of them).

The power config is set to not auto power on, the NICs are set to not WOL.

Any idea where else to look?! OS is 2008 R2.

Reset the BIOS to factory defaults. Make sure you *don't* pick the one that wipes the drives as well!!

nutcase · 8 Aug 2016 at 13:44

rotor said:
You are installing HPE drivers on a Dell server?

Other way round. All HPE servers, One Dell tape library (TL4000) and one HPE (MSL 2024).

nutcase · 8 Aug 2016 at 13:45

rotor said:
Reset the BIOS to factory defaults. Make sure you *don't* pick the one that wipes the drives as well!!

Hoping to avoid that - just in case

Will just pull the power leads if need be!

kefkef · 8 Aug 2016 at 16:30

nutcase_1uk said:
So the dying G5 won't give in. All the services are (I think) transferred off, and to test I shut it down - give it a couple of weeks and if no-one whinges about anything, wipe it (I have a backup of it anyway to retrieve any files hiding there but nowhere else).

So I turned it off. Half an hour it was back on. So I turned it off again. 2 hours later is was back on. Looking at the iLO log it always comes back on exactly on the hour (but not every hour as there was a 2 hour gap for one of them).

The power config is set to not auto power on, the NICs are set to not WOL.

Any idea where else to look?! OS is 2008 R2.

Pull the power cables out?....

Little_Crow · 9 Aug 2016 at 13:39

nutcase_1uk said:
Hoping to avoid that - just in case Will just pull the power leads if need be!

And if that doesn't fix it? you always have to be 2 moves ahead in this game.
Get googling local exorcists just in case...

swain90 · 9 Aug 2016 at 14:39

Are there any tasks in the task manager which could be causing this?

Seems rather strange, but could be a possibility?

My server turns it self on and off (technically hibernates) by using the tasks.

However if all else fails;

tribz · 9 Aug 2016 at 20:01

Spiceworks turns a lot of our kit back on, even though it's set not to!

Quartz · 9 Aug 2016 at 21:04

Don't worry about fragmentation on your servers. NTFS is resilient to fragmentation.

AHarvey · 9 Aug 2016 at 23:39

As I am in a similar position that I've moved from SEO/webwork to SEO/web/IT/Networking/backend dev and I've read through this thread, I understand none of it

Steep learning curve ahead I think.

nutcase · 10 Aug 2016 at 05:40

Ghost busters? I'm going straight for Chuck Norris

Interesting - we do have spiceworks so will have a dig around.

NTFS is more resilient, but not immune. I don't think it's been looked after very well. It's STILL 35% fragmented and the backups are running horrifically slowly.

Learning curves can be challenging but very rewarding.

ov_sjo · 10 Aug 2016 at 08:04

Sure it's not a thermal shut down?

Little_Crow · 10 Aug 2016 at 08:25

What backups are you doing that are really slow?

If it consists of lots of very small files there is likely little you can do, it's always a problem.

The way we've tackled it is to divide our Full backups over 7 days. Monday 'a-e' folders run their full, with every other day an incremental. Tuesday 'f' folders run their full, every other day an incremental, etc.

It's a lot more complicated to setup, and takes a bit of time to calculate how to divide it up somewhat evenly across the week but it's working really well for us.

nutcase · 10 Aug 2016 at 14:11

If it's a thermal issue I'd expect to see something in the logs? I think the shutdowns/restarts are down to old age, but the turning itself on again on the hour is still stumping me!

The monthly backup is literally everything on the server. Baremetal the works. The idea being restore to the latest monthly then use the dailys to bring it up to date.

I think it's the fragmentation doing it as it really is in a complete mess

nutcase · 10 Aug 2016 at 21:30

To try and clear space to help with defragging I've been archiving old user accounts - found a folder of profiles from server 2003 days.

129 folders, and at least four of the former users are dead. None in service, but 2 cancer, one old age and one murdered. Not sure why but that made me think a bit.

skyripper · 10 Aug 2016 at 21:48

Take a look in performance monitor at %disk time, which should flag up if fragmentation is a problem. Maybe hookup an external drive to backup too then leave the tape drive backing that up?

Little_Crow · 11 Aug 2016 at 08:57

nutcase_1uk said:
The monthly backup is literally everything on the server. Baremetal the works. The idea being restore to the latest monthly then use the dailys to bring it up to date.

I think it's the fragmentation doing it as it really is in a complete mess

Naturally it depends what you also have housed on C: , but any way you can just bare metal the OS Files, and then do a traditional backup for your file shares, etc?
You also want to exclude the swap file if it's not already. We sidestep the issue on our VM's by having that on it's own drive, but on a physical box it may not be feasible.

I think you need to split the backup into more manageable chunks, baremetal is good for a full DR situation, but is a bit of a sledgehammer when it comes to needing a file level restore.
What if the CEO demands a file that he spent the whole day creating and it only appears in the baremetal backup - how do you get it back?

nutcase · 11 Aug 2016 at 09:22

Average over 514 values - 1103 so sounds bad.

nutcase · 11 Aug 2016 at 09:26

Little_Crow said:
What if the CEO demands a file that he spent the whole day creating and it only appears in the baremetal backup - how do you get it back?

By using the nightly data backups. Or shadow copy.

Trials and tribulations of a new Admin.

Deleted member 138126

Deleted member 138126

Deleted member 138126

Deleted member 138126