VirtualBox

Ticket #6928 (closed defect: fixed)

Opened 4 years ago

Last modified 4 years ago

Terminal Services Guest causes CPU Storm (eats CPU)

Reported by: ToddAndMargo Owned by:
Priority: major Component: guest smp
Version: VirtualBox 3.2.6 Keywords:
Cc: Guest type: Windows
Host type: Linux

Description

Hi All,

Host: CentOS 5.5 x64 uname -r: 2.6.18-194.3.1.el5 memory: 24 GB ECC Processor: Intel x5650 (six hard cores) motherboard: Supermicro X8SAX, Bios 1.1a

Guest: Windows Server 2003, SP2, x32 running Terminal Services Guest Additions: 3.2.2

VM: VirtualBox-3.2-3.2.2_62298_rhel5-1.x86_64.rpm

Hi All,

My Guest can run Terminal Services without problem with two cores. When I increase the guest to four cores, the Terminal Server will run for a few hours and then everyone starts complaining about how slow it gets. Looking at CPU usage on the guest shows 100%.

We have Thread Master "CPUThresholdPct" set to 12% and "MainSampleTime" set to 15%. Thread Master on or off makes no difference: runs for a while, then 100% CPU usage.

I have a second customer with the same set up (12 GB, instead of 24) and Windows Server 2008 also as a Terminal Server. They are experiencing the same problem under 3.1.8. I must leave them on two cores as well.

These two customer are pretty angry at me for selling them a U$D 1200.00 processor that they can only use two cores on the guest. Please fix this soon. The natives are very restless!

Many thanks, -T

Attachments

VB-PPro2.log.tar.gz Download (64.9 KB) - added by ToddAndMargo 4 years ago.
log files for 4 cores
ws03.log.tar.gz Download (57.8 KB) - added by ToddAndMargo 4 years ago.
Windows Server 2003 freeze up
ws03-4core.log.tar.gz Download (59.2 KB) - added by ToddAndMargo 4 years ago.
3.2.4 ws2003 4 cores freeze up
ws08-4cores.log.tar.gz Download (42.8 KB) - added by ToddAndMargo 4 years ago.
Windwos Server 2008, 4 cores, 100% CPU
ws08-100cpu-performance.jpeg Download (98.3 KB) - added by ToddAndMargo 4 years ago.
WS08 100% Task Manager screenshot performance tab
ws08-100cpu-processes.jpeg Download (97.8 KB) - added by ToddAndMargo 4 years ago.
Windows Server 2008 100% CPU Processes tab -- don't add up to 100%
VB-PPro2.PerfMon.jpeg Download (162.1 KB) - added by ToddAndMargo 4 years ago.
Windows Server 2003 Performance Monitor and Task Manager
PPro2Host.top.jpeg Download (219.9 KB) - added by ToddAndMargo 4 years ago.
"top" screenshot of Linux Host for WS03 guest

Change History

Changed 4 years ago by ToddAndMargo

log files for 4 cores

comment:1 Changed 4 years ago by ToddAndMargo

Oh no. I am in *real* trouble here. Now I have two customers mad at me.

Just got this message from the second customer with the Windows Server 2008 Terminal Server. They are on 3.2.2 and two processors (out of six on their x5650). So two storms out as well: it just takes a lot longer to occur.

I am happy to report that the server has performed beautifully all morning! at one point we had nine users logged in at one time... when I went to log in, the desktop appeared as instantly as I hit the ENTER key after putting in the password!

When I came back from lunch however, I got a call from [redacted] informing me that in spite of the wonderful morning, it has become crazy: When they type, it takes a few seconds for the letters to appear, the screen gets distorted many other strange things were occurring. On our end, we are simply noticing a significant slow-down in functionality. screen freezes, click on something & you have time to go refill your coffee before it decides to react.

The customer rebooted the guest and was up and running again (same amount of users). (I was also told I am going to get a polite talking to tomorrow afternoon.)

-T

comment:2 Changed 4 years ago by ToddAndMargo

I just upgraded the server with the Windows Server 2008 guest to 3.2.4 from 3.2.2.

I tried turning off "Nested Paging": the guest would not boot past the "Microsoft" logo. Restoring Nested Paging allowed it to boot. I also put it to 4 cores as the change log said there were "stability" fixes. (I live dangerously.)

comment:3 follow-up: ↓ 4 Changed 4 years ago by sandervl73

Could you disable clipboard sharing for those VMs as a quick test? Also CPU usage of the VBoxHeadless process (for each individual thread) might be helpful.

comment:4 in reply to: ↑ 3 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

Could you disable clipboard sharing for those VMs as a quick test? Also CPU usage of the VBoxHeadless process (for each individual thread) might be helpful.

Regular CPU usage or CPU storm or both?

Also, please forgive my ignorance, but how do I do the two things you requested?

Many thanks, -T

comment:5 follow-up: ↓ 8 Changed 4 years ago by sandervl73

When the VM's performance goes down the drain. Another useful test is to take a snapshot when that happens and see if the poor performance continues after restoring the same snapshot.

As for getting CPU statistics on Linux hosts please use google. I'm not a Linux expert.

comment:6 Changed 4 years ago by ToddAndMargo

The Windows Server 2000 froze up. So I upgraded it to 3.2.4 from 3.2.2. Upgraded the guest additions as well. They are now on four cores and are happy for now. I have the log files, which I copied after the crash and before the upgrade

comment:7 Changed 4 years ago by ToddAndMargo

Sorry, that should have been Windows Server 2003, not 2000 (be nice if we had and "edit" function).

Changed 4 years ago by ToddAndMargo

Windows Server 2003 freeze up

comment:8 in reply to: ↑ 5 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

As for getting CPU statistics on Linux hosts please use google. I'm not a Linux expert.

Did not realize that is what you meant. I will use one of the Linux utilities like top.

comment:9 Changed 4 years ago by ToddAndMargo

Okay. That was quick. I fired up the ws2003 server with four cores and it completely froze up. Took about a hour. Will attach logs for that too. I put them on one core for now. We will see.

Changed 4 years ago by ToddAndMargo

3.2.4 ws2003 4 cores freeze up

comment:10 Changed 4 years ago by ToddAndMargo

Okay. The Windows Server 2008 just went to 100% on CPU. I moved them down to 1 core. I will attach logs for it.

Changed 4 years ago by ToddAndMargo

Windwos Server 2008, 4 cores, 100% CPU

Changed 4 years ago by ToddAndMargo

WS08 100% Task Manager screenshot performance tab

Changed 4 years ago by ToddAndMargo

Windows Server 2008 100% CPU Processes tab -- don't add up to 100%

comment:11 Changed 4 years ago by ToddAndMargo

The Windows Server 2003 just crashed with 1 core. I will turn off Nested Paging and VT-x tomorrow morning. I just did so on the WS2008 server

comment:12 follow-up: ↓ 13 Changed 4 years ago by sandervl73

Please do the following things as I suggested above:

  • disable clipboard sharing in the settings dialog of the VM
  • create a snapshot when the guest uses 100% cpu (but is still alive), close the vm, and restart it using that snapshot; does the high load return?

The tasklist screenshot is incomplete. Are there any processes inside the VM using a lot of cpu time?

comment:13 in reply to: ↑ 12 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

Please do the following things as I suggested above:

  • disable clipboard sharing in the settings dialog of the VM

I do not know how.

  • create a snapshot when the guest uses 100% cpu (but is still alive), close the vm, and restart it using that snapshot; does the high load return?

This is a production level server. If I wait around for that, I will probably be fired. When it crashes, they want it back up instantly and they are usually the ones that restart it, not me. If I do not get something working TODAY, I may not have a job anyway. Please ask me something a little more practical

The tasklist screenshot is incomplete. Are there any processes inside the VM using a lot of cpu time?

The "tasklist screenshot" was of the guest. I do not know of any other way of finding our guest CPU load than this.

Also, when things slow down to a crawl, the guest graphs suddenly become identical for all CPUs.

This morning, I am going to remove the $1,200.00 x5650 Westmere and replace it with an i7-920 Nehalem. This to remove Virtual Box 3.2.x's support for Intel's new Westmere VT-x code revisions and where I do believe the problems lies. I will report back.

comment:14 follow-up: ↓ 15 Changed 4 years ago by sandervl73

I very much doubt that as both CPUs are more or less the same. The only new VT-x feature is unrestricted guest execution which isn't used after early boot.

The tasklist screenshot is incomplete as there's a scrollbar.

Clipboard: VM settings, General, Advanced tab.

When the VM slows down, how does the server behave otherwise? Normal, slow? Does disabling hyperthreading help? There are also known issues with power saving modes with the new Intel CPUs. If there's a C1E setting in the BIOS, try disabling it. If there's some power saving scheme used in Linux, then turn it off.

That's about all I can do from a distance using this unofficial support channel.

comment:15 in reply to: ↑ 14 ; follow-up: ↓ 19 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

I very much doubt that as both CPUs are more or less the same. The only new VT-x feature is unrestricted guest execution which isn't used after early boot.

The tasklist screenshot is incomplete as there's a scrollbar.

Everything under the scroll bar is 0%. I arranged it in descending order. The numbers do not add up to 100%

Clipboard: VM settings, General, Advanced tab.

Right under my nose. Thank you. Clipboard is now off on the WS2003 server, which is also reverted to 3.1.8 and on the i7-920

When the VM slows down, how does the server behave otherwise? Normal, slow?

Unbearably slow. Everything seems to work, if you have an hour to wait for a 1/2 second job to complete. Sometimes it just seizes up (stops accepting network connections).

Does disabling hyperthreading help?

Not an option on this bios

There are also known issues with power saving modes with the new Intel CPUs. If there's a C1E setting in the BIOS, try disabling it.

If there's some power saving scheme used in Linux, then turn it off.

This is CentOS 5.5. You are presumed to be using it as a server, so no power saving (yipee!)

That's about all I can do from a distance using this unofficial support channel.

I have scoured Orcacle's web site looking for "official" support. I could find no such animal. Where do I go to purchase support incidents or to buy a contract? I am in serious trouble here: I could loose 3/4 on my income over this. I am very motivated to get "official" support.

Many thanks, -T

comment:16 follow-up: ↓ 17 Changed 4 years ago by sandervl73

It's kind of weird that the whole machine grinds to a halt. A runaway 4 CPU guest can't take more than 4 host CPUs. (and you should have 6 real cores and 6 hyperthreaded ones, right?) This implies something else is causing the overall slowdown.

What does top (on the host) say?

comment:17 in reply to: ↑ 16 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

It's kind of weird that the whole machine grinds to a halt. A runaway 4 CPU guest can't take more than 4 host CPUs. (and you should have 6 real cores and 6 hyperthreaded ones, right?) This

implies something else is causing the overall slowdown.

With the x5650 I have 6 cores and 12 Hyperthreads. i7-920 give me 4 cores and 8 hyperthreads.

The whole machine does not grind to a halt. I misread what you asked. My bad -- you were clear. The Guest grinds to a halt. I leave the Host with 2 cores of its own and the host is completely impervious to what is going on on the guest. When the guest storms, there is no indication of it on the host. The host works beautifully no matter what the guest is doing. (CentOS 5.5 loves those x5650 chips.)

I get the guest CPU race condition with 1,2,4 cores assigned to the guest. I have a test going on the 2008 guest (3.2.4) with one core and VT-x and Nested Paging both turned off. So far, no complains from the customer. (I wonder if he noticed it about 30% slower?)

I can not use 3 cores on the guest as M$SQL server freaks out and all h--- breaks loose.

I was also sternly told by those smarter than me on the forum not to assign more cores than you physically have -- Hyperthreads do not count when configuring a VM (and not to give all of them to the VM).

What does top (on the host) say?

I don't know as the customer restarts it before I can take a look at it. I wrote them a headless stop and start script to run from the host.

The times I have got to look at it with gnome-system-monitor have shown whatever number of processor that are assigned to host to be at 100%. The rest of the processors are at minimal.

Many thanks, -T p.s. So far (two hours) the ts2003 with 3.1.8 and the i7-920 is very stable. One lady complained that her check run was slow, but this makes sense as she was being clamped by Thread Master

p.s.s so far not fired, but got a lot of dirty looks.

comment:18 Changed 4 years ago by ToddAndMargo

Update: the ws2003 server with the replacement i7-920 and VB 3.1.8 has now been up for 10-1/2 hours without a crash. Before it would have crashed at least three time by now. The CPU hit 100% in the afternoon, but the server held steady and I was able to operate anything I wanted (a bit slowly).

I did adjust the clamp (Thread Master) to 3 seconds at over 8% gets you clamped to 3% CPU. This calmed the 100% on the CPU: made everyone happy. (Although the people running the offending reports did take a bit longer.)

I do believe that my U$D 1,000.00 x5650 in not compatible with Virtual Box at this juncture.

-T

comment:19 in reply to: ↑ 15 ; follow-up: ↓ 20 Changed 4 years ago by michael

Replying to ToddAndMargo:

I have scoured Orcacle's web site looking for "official" support. I could find no such animal. Where do I go to purchase support incidents or to buy a contract? I am in serious trouble here: I could loose 3/4 on my income over this. I am very motivated to get "official" support.

You might want to take a look at  http://forums.virtualbox.org/viewtopic.php?f=16&t=31864.

comment:20 in reply to: ↑ 19 ; follow-up: ↓ 21 Changed 4 years ago by ToddAndMargo

Replying to michael:

Replying to ToddAndMargo:

I have scoured Orcacle's web site looking for "official" support. I could find no such animal. Where do I go to purchase support incidents or to buy a contract? I am in serious trouble here: I could loose 3/4 on my income over this. I am very motivated to get "official" support.

You might want to take a look at  http://forums.virtualbox.org/viewtopic.php?f=16&t=31864.

Hi Michael,

There is bug in the forums that gives me:"You are not authorised to read this forum" even though I am logged in and post all the time on that forum. It seems to be all the old post that give me this problem.

Is there a way you an copy and past the information and eMail it to me?

Many thanks, -T

comment:21 in reply to: ↑ 20 ; follow-ups: ↓ 23 ↓ 24 Changed 4 years ago by michael

Replying to ToddAndMargo:

There is bug in the forums that gives me:"You are not authorised to read this forum" even though I am logged in and post all the time on that forum. It seems to be all the old post that give me this problem.

Is there a way you an copy and past the information and eMail it to me?

"Re: Commercial Lincense by fatbloke » Thu Jun 10, 2010 9:36 am

Commercial licenses are available from Oracle, although at the moment you have to purchase from Oracle sales or an Oracle partner. The part number to order is: XVBII-LCO-9929 and is called a VirtualBox per User Perpetual License Entitlement. (Roughly $50 list and you can ask for Premier Support with that)

We are also hoping to make this available on the Web Store, but it may take some time

  • FB"

comment:22 Changed 4 years ago by ToddAndMargo

update. Both servers are stable all day today.

comment:23 in reply to: ↑ 21 Changed 4 years ago by ToddAndMargo

Replying to michael:

Commercial licenses are available from Oracle, although at the moment you have to purchase from Oracle sales or an Oracle partner. The part number to order is: XVBII-LCO-9929 and is called a VirtualBox per User Perpetual License Entitlement. (Roughly $50 list and you can ask for Premier Support with that)

Thank you! I will give them a call. (I hope it comes with direct support.)

-T

comment:24 in reply to: ↑ 21 ; follow-up: ↓ 26 Changed 4 years ago by ToddAndMargo

"Re: Commercial Lincense by fatbloke » Thu Jun 10, 2010 9:36 am

Commercial licenses are available from Oracle, although at the moment you have to purchase from Oracle sales or an Oracle partner. The part number to order is: XVBII-LCO-9929 and is called a VirtualBox per User Perpetual License Entitlement. (Roughly $50 list and you can ask for Premier Support with that)

FB,

Okay. Over two days, I have tried calling Oracle. The right hand does not know what the left hard is doing. I left off with your marketing rep will "reach out" to you. All marketing psychobabble aside, I have had no luck. I tried. I really tried. Do you have a phone number I can call?

-T

comment:25 in reply to: ↑ description Changed 4 years ago by badboi

Hi!

I got into the same problem. host: Windows XP 64bit, 4 cores, 4GB ram guest: Windows Xp 32bit, 1 GB ram newsest virtualbox (3.2.4 rc52467)

If I assign 4 cores: -> guest crash or 100% CPU (not able to access guest anymore->shutdown) if I assign 2 cores -> no problem I can see

I can trigger the problem by starting 2 huge file transfers from host to guest (many files). Same efect if I use networkpath or shared directories. If I do this after around 2-5 minutes crash or 100% CPU

crash also occured when I downloaded a file from internet

something eles? *) I tried nested paging on and off.. no difference *) on the same host another guest, same settings, but one core.. runs smoothly since months

Regards Michael

comment:26 in reply to: ↑ 24 ; follow-up: ↓ 27 Changed 4 years ago by michael

Replying to ToddAndMargo:

Do you have a phone number I can call?

I'm sorry, but what I gave you is all the information I have. Sadly (or perhaps happily) as an engineer I don't have much contact with the sales and marketing side of things.

comment:27 in reply to: ↑ 26 Changed 4 years ago by ToddAndMargo

Replying to michael:

Replying to ToddAndMargo:

Do you have a phone number I can call?

I'm sorry, but what I gave you is all the information I have. Sadly (or perhaps happily) as an engineer I don't have much contact with the sales and marketing side of things.

After two weeks of calling, Oracle finally eMailed me:

I understand from our marketing team that you had some questions regarding VirtualBox. Sounds like you may have the software today and were interested in support. I reached out to our production team to better understand the support model for the open source version vs the full use version. It appears the full use version product is not ready yet for release and support by Oracle. Looks like this is still a month or two out. The product team recommended using the open source version for now and they hoped to have more details in July/Aug. Please let me know if you have any additional questions in the meantime.

comment:28 Changed 4 years ago by ToddAndMargo

Hi All,

Since I can not purchase support, would you guys escalate this as if I could. I am in real trouble here.

Many thanks, -T

comment:29 Changed 4 years ago by ToddAndMargo

Okay, apparently this bug applies to Windows hosts as well:  http://forums.virtualbox.org/viewtopic.php?f=6&t=31836&p=145707#p141862

Changed 4 years ago by ToddAndMargo

Windows Server 2003 Performance Monitor and Task Manager

comment:30 Changed 4 years ago by ToddAndMargo

I just attached a screen shot of the Windows Server 2003 Performance Monitor along side the Task Manager. Again, the numbers (sorted descending by CPU usage) to not add up to the CPU usage.

comment:31 Changed 4 years ago by ToddAndMargo

Hi All,

Please respond to this question.

On the Windows Server 2003, I am scheduled to replace the heat sink next Thursday morning. At that time, it would be easy to remove the current i7-920 processor and replace it with the original X5650 processor.

Will the X5650 be just as stable as the i7-920 under VB 3.1.8?

If not, please stop me. I could get in real trouble. I really need you guys to respond to this question.

Many thanks, -T

Changed 4 years ago by ToddAndMargo

"top" screenshot of Linux Host for WS03 guest

comment:32 Changed 4 years ago by ToddAndMargo

I just attached a "top" screen shot of the host (CentOS 5.5 x64) that is running the Windows Server 2003 Terminal Server. You will notice the first Headless is the WS2003 guest. The second well behaved Headless is Windows XP Pro.

Only the first headless (ws03 guest) is showing any CPU activity in the slightest.

-T

comment:33 Changed 4 years ago by ToddAndMargo

Moderator: please change the title of this bug to:

Terminal Services Guest causes CPU Storm (eats CPU)

Many thanks, -T

comment:34 Changed 4 years ago by michael

  • Version changed from VirtualBox 3.2.2 to VirtualBox 3.2.6
  • Summary changed from 2 cores are stable, 4 cores eventually gives CPU storm to Terminal Services Guest causes CPU Storm (eats CPU)

comment:35 Changed 4 years ago by ToddAndMargo

Hi All,

As of 07/16/2010, I replaced the heat sink on the Windows Terminal Server 2003. I took advantage of the situation and removed the i7-920 and reinstalled it with the original x5650. I set the VM (3.1.8) to one CPU (core) and turned VT-x off (nested Paging is still on). The Terminal Server is now completely stable, works better that it did with two (i7-920) cores on VT-x, and there are no more CPU storms.

Please note that the other Windows Terminal Server 2008 was not stable on one core with VT-x enabled. It only stabilized when VT-x was disabled.

I hope this helps narrow down the problem, which I do believe is VT-x. Let me know if you have anything to test.

Many thanks, -T

comment:36 Changed 4 years ago by ToddAndMargo

I am not seeing anything in 3.2.8 that addresses this. Am I correct?

comment:37 Changed 4 years ago by ToddAndMargo

Update. I upgraded one of the Terminal Server to VBox 3.2.8 from 3.2.4. Did not help. The Terminal Server (one core, software emulation) ran about 4 times slower than under 3.2.4. Customer kind of freaked out. So I downgraded him to the revision the other customer is using: 3.1.8. Happy camping has resumed. 3.1.8 is a bit faster than 3.2.4 too.

I really, really hope you guys are working on this. I really need this fixed.

Many thanks, -T

comment:38 Changed 4 years ago by ToddAndMargo

Any progress?

comment:39 Changed 4 years ago by sandervl73

  • Component changed from other to guest smp

comment:40 follow-up: ↓ 42 Changed 4 years ago by sandervl73

It might be solved by 3.2.10 if you enable large page support for the VM. We've received a similar report from a customer which has confirmed it's fixed with the latest internal build. You can't enable large page support in 3.2.8 as there's a bug that might trigger a guru meditation.

3.2.10 is due in the next few weeks if all goes well.

comment:41 follow-up: ↓ 43 Changed 4 years ago by sandervl73

You can also try the following workaround with 3.2.8:

VBoxManage setextradata "VM name" "VBoxInternal/PGM/Pool/MaxPages" 4096

comment:42 in reply to: ↑ 40 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

It might be solved by 3.2.10 if you enable large page support for the VM. We've received a similar report from a customer which has confirmed it's fixed with the latest internal build. You can't enable large page support in 3.2.8 as there's a bug that might trigger a guru meditation.

3.2.10 is due in the next few weeks if all goes well.

Very cool. Thank you.

Is there any technical description of the problem and its solution you could post here?

comment:43 in reply to: ↑ 41 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

You can also try the following workaround with 3.2.8:

VBoxManage setextradata "VM name" "VBoxInternal/PGM/Pool/MaxPages" 4096

Way too chicken. Both these serer are full production level servers. I have to be very careful of outages. I will wait till .10 appears.

comment:44 follow-up: ↓ 45 Changed 4 years ago by sandervl73

3.2.10 will do the same as that VBoxManage command, but automatically based on the guest ram size. That fixes performance issues such as yours for SMP guests with large ram configs.

The actual description of the problem is rather technical.

comment:45 in reply to: ↑ 44 Changed 4 years ago by ToddAndMargo

Replying to sandervl73:

3.2.10 will do the same as that VBoxManage command, but automatically based on the guest ram size. That fixes performance issues such as yours for SMP guests with large ram configs.

The actual description of the problem is rather technical.

Hi Sandervl73,

The problem exists with a single processor and hardware emulation. Does it address the "CPU Storm" problem with hardware emulation? The problem is the 100% CPU usage after a bit of actual run time. It does not "storm" with a single processor and software emulation.

Also, does it is correct bug #6936, which I do believe is a duplicate of my bug?

Many thanks, -T

comment:46 Changed 4 years ago by sandervl73

#6936 sounds different.

The problem tends to show up with nested paging, SMP and large guest ram configs only, but SMP is not a requirement. Guest SMP just is much more likely to show the problem.

comment:47 Changed 4 years ago by sandervl73

  • Status changed from new to closed
  • Resolution set to fixed

comment:48 Changed 4 years ago by ToddAndMargo

Got the server with the ws03 VM upgraded from 3.1.8 to 3.2.10. Changed it from one soft processor to four hardware processors. It is working. I will hold off a week on the second server with the WS08 guest, as this customers looses his customers when the server goes down.

Found something interesting. Both Tops on the Linux host and "Task Manager" on the guest showed very low processor usage. But "Performance Monitor" on the guest showed a constant 25% usage after all the users logged into the guest's Terminal Server. Before they logged in Performance Monitor's CPU usage was essentially zero. After a little research, I found that "Performance Monitor" is showing an "average" over one second of usage. Telling Tops to update every 0.2 second showed the up to 80% CPU fraction of a second hits that did not show up in Tops with a higher sampling rate. Also both Tops and Task Manager are taking snap shots and not averages. So all was okay after all.

Thank you guys so much for fixing this! You can not imagine! -T

comment:49 follow-up: ↓ 50 Changed 4 years ago by ToddAndMargo

Okay, I got brave and upgraded the Windows Terminal Server 2008. The CPU storm is gone. But, the massive slowdown I reported on 2010-09-15 when I tried to upgrade from 3.2.4 to 3.2.8 is still in effect. A customer got his database corrupted in the process of testing it too. So, I had to revert to a single soft processor and 3.1.8. I restored the corrupted database to Saturday night's snapshot. (Fortunately, they did not work on it on Sunday).

So, this bug really is fixed. I just have another Windows Server 2008 specific bug to deal with and I will report that under another bug.

I am one for two.

Thank you all for all the help.

-T

comment:50 in reply to: ↑ 49 Changed 4 years ago by ToddAndMargo

Replying to ToddAndMargo:

I just have another Windows Server 2008 specific bug to deal with and I will report that under another bug.

new bug is 7607

Note: See TracTickets for help on using tickets.

www.oracle.com
ContactPrivacy policyTerms of Use