RM CC3 Rebuilds Failing

[Hightower]

This is bizarre...

 

I tried to rebuild an RM machine with an E1000 NIC in, which kept failing - no particular error message, it just seemed to hang when it got to the "Copying installation files (this will take a few minutes)...." screen.

 

So, using the exact same floppy I went to another room and successfully built two computers straight away, with a VIA and a SIS NIC - again, both RM machines. Then I went to build a machine with a Gig 8169 (again, RM machines) and it too stopped at the "Copying installation files" screen.

 

I'm not sure whether the NIC's is the cause, just posting them incase it is relevant.

 

So I tried recreating the disk - didn't work

Then I tried recreating the type with only the 8169 and E1000 drivers - didn't work

So I copied the RMRebuild files on SVR-001 (as a backup), deleted them and copied SVR-002 files into the SVR-001 directory. Recreated disks and types - didn't work.

 

The bizarre thing is I built 2 computers fine last night using the very same disk.

 

I've managed to get one away by installing a Vanilla XP and Smart Clienting it, but I want to hold on to the other one until i get this working.

 

Any help?

[CommodoreS]

What is the latest Hardware Support Pack installed on your network?

 

Have you installed any new HSPs recently, and is it only since they were installed the machines will not rebuild?

 

I would also take the HSPs to the latest version if the latest is not installed. Some HSPs fix problems that were created by older HSPs (but do not always tell you this in the notes!). At the moment installing up to the most recent HSP seems quite stable.

 

The Build Disks have changed a bit recently, as they have got too big to hold all the network card drivers. As a result they are now split into two, and (from memory) to get some to work I think you have to run an extra tool.

 

The only other thing I would double check the cards are E1000 as the autodetect on the build disks can sometimes mis-report what the NIC actually is and that can cause the build to fall over.

 

It also looks like someone on RM Communities has the same problem as you, so it may be worth checking their post: RM Communities

[bossman]

@Hightower:

 

Hiya bud, do a memtest on the failing machines and you probably find it is this that maybe the problem. Also check the HDD as this also has been a problem in the past. Last but not least it could be the motherboard as we discovered this on one workstation which failed for the exact reason you have described.

 

these are the only times we have found that the build process fails.

 

Hows me friend Margaret doin? :)

[sparkeh]

it just seemed to hang when it got to the "Copying installation files (this will take a few minutes)...." screen.

 

Through a lot of testing I believe that this comes down to a motherboard issue, mostly a chipset issue.

 

Its actually a really weird problem as if you stop the build process and look through the scripts to see what it is doing, it appears that its only contacting the build server to get the correct files. Through trying to build 3rd party machines I found a much greater failure rate with machines with cheaper chipsets (or motherboard issues) but can't for the life of me work out why.

[CommodoreS]

Through a lot of testing I believe that this comes down to a motherboard issue, mostly a chipset issue.

 

Its actually a really weird problem as if you stop the build process and look through the scripts to see what it is doing, it appears that its only contacting the build server to get the correct files. Through trying to build 3rd party machines I found a much greater failure rate with machines with cheaper chipsets (or motherboard issues) but can't for the life of me work out why.

 

Does not sound like this to me with it being 3 different computers and 2 different chipsets. It could be a network infrastructure issue (spanning tree etc...), althouth that seems unlikely. The way to test that would be to try building one of these three machines directly out the back of server 1 using a cross over cable.

 

There is a problem with some AIOs and the memory becoming loose. The way to test this is to take the machine to the logon screen and hit the RM AIO on the top quite hard with your hand (try not to hurt your hand). If the machine then reboots the memory has become loose inside. Unscrew the machine, remove the RAM modules and reinsert them and this fixes this unusual problem.

[Hightower]

Thanks for the responses

 

Firstly, we're up to HSP 39 on all CC3 servers - I believe this is the latest HSP.

 

Secondly, the person experiencing the same problems on Communities is..... me! So unless I develop schizophrenia I doubt I'll be able to help myself

 

Next up, I've tried running PC Check for 48 hours on the failing machines and all components passed with flying colours.

 

Finally, these machines have both been built numerous times without fail, so obviously something has changed somewhere for this to start happening. The problem is, I haven't built these machines (or the same models) in some time so I can't pinpoint exactly when these symptoms started.

[TechMonkey]

Would they build off a memory stick? I think we had a similar problem but the difference was we were using a memory stick and the machine didn't like usb floppy & wanted USB HD.

[Hightower]

Would they build off a memory stick? I think we had a similar problem but the difference was we were using a memory stick and the machine didn't like usb floppy & wanted USB HD.

 

No - I tried a memory stick the first time, and this didn't work. I initially thought this was the problem and so tried using an external floppy (the one that worked in another room) but this didn't work either.

[Hightower]

UPDATE: Just tried reseating the RAM and set away again. This is the error message received this time:

 

Copying installation files (This will take a few minutes)...
CIADV.MS_

Extended Error 53

 

I also got one earlier today on the other machines (but failed to mention - sorry) like this:

 

Copying installation files (This will take a few minutes)...
ICWCONN1.EX_

Extended Error 53

[webman]

Check hard disk cables are OK?

[Hightower]

Check hard disk cables are OK?

 

Just tried with new IDE cable - still no go.

[bossman]

I would be looking at another motherboard.

 

I would swap all components except motherboards out of another identical workstation if you have one spare and see what you get.

[Hightower]

I would be looking at another motherboard.

 

I would swap all components except motherboards out of another identical workstation if you have one spare and see what you get.

 

Just seems a bit coincidental to me that this is happening on 2 completely different spec machines at the same time - I would have thought it was more something on the network than an individual hardware failure....

[CommodoreS]

Just seems a bit coincidental to me that this is happening on 2 completely different spec machines at the same time - I would have thought it was more something on the network than an individual hardware failure....

 

Definition of: Extended Error 53. A DOS error message that typically means a network connection has been broken between two stations.

 

I would now try with a crossover cable directly from one of the machines to the back of server1 to see if there is network infrastructure issue (and also to rule this out as a possibility).

 

There is a remote possibility it might be bad hard drives. With Bad Sectors these normally fall over when trying to create a RM Restore Partition, and you end up with a machine with no RM Restore Partition. There was a problem with some SATA cables. There is also a program you can use to check this and is available with instrucstions from TEC1049164 on the RM Support Website.

[Hightower]

I'm running another PC Check through the machine to double check.

 

I've just stumbled on something interesting with DNS though and am waiting for the LEA to contact me to see if this could be the cause?

[Hightower]

DNS wasn't the problem - any more ideas guys? PC Check is so far clean but gonna let it run over night.

[CommodoreS]

Have you tried building it on a crossover cable to rule out network infrastructure?

[Hightower]

Nope - don't have one to hand.

 

The shared tech is taking the machine tomorrow to a primary - he's going to try and build it there, which will inevitably show us whether it is hardware or network.

 

I'll keep you posted

[Sylv3r]

You could always give the server a reboot )

 

Others have mentioned issues with rebuilding to be infrastructure related rather than the software.

 

Busy period on the network when trying to rebuild - start of lessons etc? Intense piece of software being ran in the room at the time?

 

Does seem quite odd mind. Generally the build process works flawlessly.

[Hightower]

You could always give the server a reboot )

 

Others have mentioned issues with rebuilding to be infrastructure related rather than the software.

 

Busy period on the network when trying to rebuild - start of lessons etc? Intense piece of software being ran in the room at the time?

 

Does seem quite odd mind. Generally the build process works flawlessly.

 

Servers were rebooted. I know what you mean about 'busy' times, but this machine in particular was set away the same time as the three that succeeded which makes it even stranger.

 

Will update when the shared tech brings me back the results of our little experiment.

[Hightower]

Shared tech has just rang to say as soon as he set it away at a primary it failed with the partioning part of the build. Looks like a HDD problem.

 

Good to know that it's NOT our network - just a bit bamboozed how it passed all the PC Check tests :S

 

Oh well, never mind - thanks for all the help!

[webman]

@Hightower:

 

Also check the HDD as this also has been a problem in the past.

 

He's not as daft as he looks - honest!

[Hightower]

He's not as daft as he looks - honest!

 

I did check the HDD though

[webman]

It's tricksed yoouu

[Hightower]

Drat, drat, and double drat!