Server locks up every morning

[Ayman Altounji]

I just installed a new ML-370 at a customer's site. We configured it in our office with no problems over several weeks. We installed it onsite Friday and worked on it all day Friday, Saturday, and Sunday without a hiccup.

Monday morning, the customer reported that no one was able to log on. I had him try pinging the server and the Internet gateway from a Windows 98 workstation, but did not get a response from either. The server was completely locked up -- no response to mouse or keyboard. We power-cycled the server, it started normally, and worked for a while -- then locked up again around 10:00 or 11:00. After another restart, it ran for the rest of the day without problems. However, this morning it was locked up again.

The server is an ML-370, PIII 733 MHz, 128MB RAM, 4x18GB hard drives w/ no RAID card (mirroring C:, D:, and E: partitions using W2K's fault tolerance). The internal 20/40 GB DAT drive has not been configured yet. The onboard NIC is connected to an Intel 460t 24-port Ethernet switch. Power protection is provided by a new MinuteMan 420 UPS, but a UPS interface cable has not been connected.

[Ayman Altounji]

Follow up to yesterday's message -- the server was locked up again this morning. The clock showed 1:23 AM. After it was restarted, I checked the System Event Log and found there were no events in the log between 1:12:46 and 8:15:46 AM this morning (the server was restarted a few minutes after 8:00 AM). This morning, however, the mouse moved and the Caps, Scroll, and Num Lock lights did toggle when the respective key was pressed. The server also responded to pings by IP address and by name from the Proxy server.

I checked the Perf Mon log I started last night. There was nothing strange about the parameters I was logging at either 1:12 AM or 1:23 AM -- Total Processor % was between 0.5% and 2.5%, and IP Datagrams/second were between 5 and 30 all night long, with no particular peaks on either parameter.

--spurlock--

[Ayman Altounji]

Oops -- I just realized the logs are in GMT, not Central Standard Time. The log entries stopped at 7:23 AM GMT, or 1:23 AM -- the same time frozen on the Taskbar clock. Looking at the Perf Mon data again, I can see that at 1:00 AM, the Processor Interrupts/second jump from around 75 to over 300, and CPU utilization jumps from 0.5-2.5% to 23%. At 1:06, the CPU utilization dropped back to between 2 and 5%, but the Interrupts/second stayed at over 300 right up to where the log ended at 1:23 AM.

The server has frozen up two more times this morning since 9:00 AM. I did not have monitoring configured when it did, but I'm going to configure my Perf log to restart whenever the server does.

--spurlock--

[Ayman Altounji]

I am getting a similar problem with one of our servers. Does the system console screen stay blue (not Blue Screen of Death) without displaying the "Press Carl At ... etc " screen?

I am trying to isolate what software is causing it but constant monitoring is showing nothing exceptional.....

[Ayman Altounji]

Er...that should of course say "Ctrl, Alt", spell checking thingy got in the way!!!

[Ayman Altounji]

I've been looking through Compaq and Microsoft for a hint about this and have come across this in Microsoft's Kb - see Q263627 Computer May Hang with Object Auditing Enabled.

It appears to be a post-SP1 fix (or pre-SP2 depending how you read it!!).

There are also issues relating to the SCSI port locking Win2K, Q275620

Try these.

Matthew

[Ayman Altounji]

Light at the end of the tunnel? After multiple calls to Compaq and Microsoft and replacement of the motherboard and RAM, I tracked the trigger overnight as being due to Exchange maintenance firing off at 1:00 AM. I reinstalled Exchange last Friday, but that didn't make a difference.

However, I think I may have traced the problem to a problem with one of the hard drives used for the mirrored D: partition. I used Exchange Optimizer to move the Private database from D: to E:, then tried running the Exchange maintenance at the next possible quarter-hour. It took over an hour, but the maintenance completed for the first time since the Private database was created a month ago. I scheduled another online maintenance for this morning, and the server has not crashed yet.

While I'm on my soapbox, I would like to say that I am very disappointed in the apparent lack of response from Compaq technicians to my original postings in this forum. I would expect that a company that hosts such a forum on their Website would have their techs review the postings, even if the best suggestion is "Call Microsoft."

[Ayman Altounji]

Right on spur lock. I've run across several posts with my exact problem but none of them have definitive responses from Compaq. My only choice is to call which costs me time and Compaq money. I don't get it.

[Ayman Altounji]

Spurlock, "You the Man", we are having the exact same problem as you, same server, same OS and Exchange 5.5. We are going to try and move our Priv data base using Exchange Optimizer. The one difference we have compared to you is we don't have mirroring on the drives. As well I agree with your critisim on both support counts. Also are you running all of the Compaq Management tools, we are wondering if that was an issue as well?By the way had the same time issue with Pref Monitor and GMT.If this works you have saved me alot of grief. Thanks.

[Ayman Altounji]

Hi all

We are also having the same problem. I am keen to know from Johnny_Mac what happened as we are not using any Raid at the moment, and if it was sorted out We are also running a Compaq ML370, Dual I III 866 256 BB RAM 2X 18.2 SCSI He's and Win2k Server and Exchange 5.5 SP3 I will also try and move the Priv.edb and see what happens