Re: CPU usage goes way up on 5.1 when memory is added
On Sat, 24 Jan 2004 03:22:51 -0600, Quintin Riis wrote Install GENERIC kernel and see if the problem persists. Quintin I played around with various options this weekend. First I tried just switching the scheduler to the 4BSD version and got this: cpuid = 0; apic id = 00 instruction pointer = 0x0:0xa00 stack pointer = 0x0:0xffe frame pointer = 0x0:0x0 code segment= base 0x0, limit 0x0, type 0x0 = DPL 0, pres 0, def32 0, gran 0 processor eflags= interrupt enabled, vm86, IOPL = 0 current process = 0 () trap number = 30 panic: unknown/reserved trap cpuid = 0; Uptime: 1s So, I tried booting the GENERIC kernel that came with the system (even though its not SMP). This booted OK, but hung trying to start up squid and I suspect various other utilities. I'm guessing this was due to memory shortage (which was why I added the memory). There were no errors at all, so I really don't know why it got scrogged. So, then I tried taking the generic kernel conf file and adding just SMP, IPSEC IPFIREWALL support. Again, this is using the 4BSD scheduler. I got the same panic. This is all with the same 384MB of RAM that works just fine with the SMP kernel I compiled with the ULE scheduler. This system has never been so fragile, so I'm wondering if a strange hardware problem is creeping in. I just don't know why its so selective about showing its teeth though. So, I'm back to happily running the ULE SMP kernel with 384MB of memory, wishing I could just add some more memory. If anyone can offer any more insight as to why the additional RAM causes the CPU usage to flake out, I'd be most grateful. Cheers, rgr -- Richard G. Roberto [EMAIL PROTECTED] ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: CPU usage goes way up on 5.1 when memory is added
Install GENERIC kernel and see if the problem persists. Quintin Richard G. Roberto wrote: Hi, I've been running 5.1-RELEASE for quite some time on my dual PIII 333Mhz Micron PC with 384MB of RAM and 670MB of swap. Its a little low on memory for what I'm now doing with it though, so I added 256MB of RAM. The new RAM gets recognized, and there are no obvious errors on the console or in the logs, but every little thing drives the CPU utilization through the roof and the system is unusable. Even just typing man ls triggers it, never mind trying to run ssh. I remove the RAM, and it goes back to normal. I'm using a kernel with these relevent settings: cpu I686_CPU options SCHED_ULE options COMPAT_43 options COMPAT_FREEBSD4 options _KPOSIX_PRIORITY_SCHEDULING options SMP options APIC_IO options LAZY_SWITCH There are other settings, but these are the only ones that might be related, I'm guessing. Has anyone seen anything similar to this? I've searched google as well as the mailing list archives and can't really find anything similar. Any help would be greatly appreciated. Thanks in advance, rgr ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: CPU usage goes way up on 5.1 when memory is added
On Fri, Jan 23, 2004 at 06:07:20PM -0500, Richard G. Roberto wrote: Hi, I've been running 5.1-RELEASE for quite some time on my dual PIII 333Mhz Micron PC with 384MB of RAM and 670MB of swap. Its a little low on memory for what I'm now doing with it though, so I added 256MB of RAM. The new RAM gets recognized, and there are no obvious errors on the console or in the logs, but every little thing drives the CPU utilization through the roof and the system is unusable. Even just typing man ls triggers it, never mind trying to run ssh. I remove the RAM, and it goes back to normal. How are you trying to measure this? Have you tried reverting to a GENERIC kernel to isolate one of your non-default options that may be causing the problem (or measurement artefact)? Kris pgp0.pgp Description: PGP signature
Re: CPU usage goes way up on 5.1 when memory is added
On Fri, 23 Jan 2004 15:54:44 -0800, Kris Kennaway wrote How are you trying to measure this? Have you tried reverting to a GENERIC kernel to isolate one of your non-default options that may be causing the problem (or measurement artefact)? Kris Well, measuring it isn't really an issue as the problem is so drastic, there's no trouble in identifying it. I have taken some vmstat and iostat data over time that can be used for comparison, but it really isn't necessary in this case. The CPU usage isn't even close to normal with the additional memory. Since this system is in heavy use, I haven't had the time to do a lot of trial and error with different kernel options yet. I will certainly try to schedule that soon. I was hoping that somebody on the list would have an idea as to why this happens. I really can't form a relationship in my head between the additional RAM and the strange CPU behavior. So I was hoping someone more familiar with freebsd 5.1 could help me out. I know that the ULE scheduler is still experimental, so that was going to be my first choice for rolling back, but I'm still interested in hearing from anyone else who may have seen similar behavior. I'm trying to get this sorted out before my linux zealot friends harass me to no end ;-) Thanks again, rgr -- Richard G. Roberto [EMAIL PROTECTED] ___ [EMAIL PROTECTED] mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]