Latest vmware-server crashes a lot with 'double fault' errors. Complete machine breaks etc. I didn't have these issues at the previous version of this product. Reproducible: Always Steps to Reproduce: 1. emulate something in vmware 2. wait it to crash Actual Results: alot of the first three lines May 22 01:09:18 nemesis rtc: lost some interrupts at 2048Hz. May 22 01:09:18 nemesis rtc: lost some interrupts at 2048Hz. May 22 01:09:18 nemesis rtc: lost some interrupts at 2048Hz. May 22 01:09:18 nemesis /dev/vmmon[14000]: host clock rate change request 1043 -> 0 May 22 01:09:18 nemesis rtc: lost some interrupts at 2048Hz. May 22 01:09:18 nemesis vmmon: Had to deallocate locked 34951 pages from vm driver ffff810005e7e000 May 22 01:09:18 nemesis vmmon: Had to deallocate AWE 4854 pages from vm driver ffff810005e7e000 May 22 01:09:18 nemesis device eth1 left promiscuous mode May 22 01:09:18 nemesis bridge-eth1: disabled promiscuous mode When I mean crash, I mean crash, this kills the entire system eventually.
Hi Stefan, Could you please describe what you mean by "double fault" errors, from the log messages that you've posted I can't see any mention of double faults or similar. I've looked around for mention of vmmon having to deallocate space and those show up as memory issues (please check how much memory you're allocating your virtual machines). As to the rtc clock losing cycles, that can be dealt with by adding: host.useFastClock = FALSE into /etc/vmware/config. For more information have a look at http://www.vmware.com/pdf/vmware_timekeeping.pdf. Also, it's important to know which kernel it is you're using as 2.6.18 is about the last version that vmware-server was reported to be very solid on. Other than those general pieces of advice, your best bet is to ask on the VMTN or lodge a support request if you've got a support contract with vmware. They're unlikely to support Gentoo given that there are far more possible variations of the base system than with binary distributions, but I'm not certain how else we can help. I've looked through the VMTN and haven't found many other people reporting the issues you've experienced, and no one else on Gentoo has reported similar issues, so I'm afraid this looks like it might be best handled upstream. For the time being, I'll leave the bug as TEST-REQUEST whilst you try out those three suggestions above (change the memory settings, add the line to the config file, change the kernel down to 2.6.18) and we'll see how you get on...
(In reply to comment #1) > Hi Stefan, > > Could you please describe what you mean by "double fault" errors, from the log > messages that you've posted I can't see any mention of double faults or > similar. Is the gui-tool, before it crashes, I get a popup with a long message what went wrong and what to submit to VMWare. When I try to go to that page it seems you cannot submit anything. > I've looked around for mention of vmmon having to deallocate space > and those show up as memory issues (please check how much memory you're > allocating your virtual machines). As to the rtc clock losing cycles, that > can be dealt with by adding: > > host.useFastClock = FALSE > > into /etc/vmware/config. For more information have a look at > http://www.vmware.com/pdf/vmware_timekeeping.pdf. I allocated around 256MB of memory to WinXP. The same crash can I reproduce with the BasicCD / VMWare Disk of Syllable (the operationsystem). Instant crash. > Also, it's important to know which kernel it is you're using as 2.6.18 is > about the last version that vmware-server was reported to be very solid on. I'm on 2.6.21 now, but I didn't have any problems on 2.6.20 with the last version of VMware. > For the time being, I'll leave the bug as TEST-REQUEST whilst you try out > those three suggestions above (change the memory settings, add the line to the > config file, change the kernel down to 2.6.18) and we'll see how you get on... I'm going to give it a shot :) And otherwise... back to the 'stable' open sources virtualizers...