Out of curiosity, what does CPU usage look like when you start seeing > problems? How about interrupts? (To find this information, run 'vmstat > 1'--if you'd like to send me about 100 lines of the output from this via > private mail next time you begin to hit this limit, I might be able to > spot other problems that would lead to the trouble you're seeing.)
I will do this the next time. Thank you very much. Best regards, Edward Millington. BSc, Network+ Systems Administrator Cariaccess Communications Ltd. Palm Plaza Wildey St. Michael Barbados 1-246-430-7435 Fax : 1-246-431-0170 [EMAIL PROTECTED] www.cariaccess.com ----- Original Message ----- From: "Joe Cooper" <[EMAIL PROTECTED]> To: <[EMAIL PROTECTED]> Sent: Friday, March 08, 2002 10:13 AM Subject: Re: [OOPS] increasing the number of threads on linux > Some of those are extremely high, and probably cause negative impact > rather than positive. But they shouldn't cause a 1000 thread limit, > that I can see. > > I doubt RLIMIT_NOFILE is a problem. 8196 is plenty (probably a lot more > than plenty). > > I think I would lower some of those values to something a little more > reasonable, just to make sure the kernel doesn't do something > pathological when configured with such extreme limits. I would go for a > startup script something like this: > > ulimit -HSn 8192 > echo 8192 > /proc/sys/kernel/threads-max > echo 8192 > /proc/sys/fs/file-max > echo 1024 32768 > /proc/sys/net/ipv4/ip_local_port_range > echo 3072 > /proc/sys/net/ipv4/tcp_max_syn_backlog > echo 1 > /proc/sys/net/ipv4/ip_forward > > Leave out the other stuff, as it is probably counter productive or not > useful in this situation (the other ulimits are already the defaults, > why set them explicitly?). Even the kernel threads-max value at 8192 is > probably too high--you've only got 1000 right now, and probably only > 30-50 other processes on the machine. You aren't going to hit the > default 4079 limit, ever. > > I don't know that this will fix your problems. Linux may still have > problems with a huge number of threads, or Oops may not be able to > sustain the kind of traffic you're seeing--and it just reaches that > limit at a certain client load. Threads aren't exactly the ideal way to > handle a large client population--and this may be the limit of the Linux > threading implementation. It would be a kernel bug if it is, but I just > don't know how much the threads implementation gets pushed up over 1000 > threads in common usage. > > Out of curiosity, what does CPU usage look like when you start seeing > problems? How about interrupts? (To find this information, run 'vmstat > 1'--if you'd like to send me about 100 lines of the output from this via > private mail next time you begin to hit this limit, I might be able to > spot other problems that would lead to the trouble you're seeing.) > > Edward Millington wrote: > > my socket was default under squid. 8192. Now set to 12288 > > > > Here is my default config. for over3 months. > > > > ulimit -HSn 12288 > > ulimit -HS -d unlimited > > ulimit -HS -s unlimited > > ulimit -HS -c unlimited > > ulimit -n 32768 > > echo 100000 > /proc/sys/kernel/threads-max > > echo 32768 > /proc/sys/fs/file-max > > echo 4096 32768 > /proc/sys/net/ipv4/ip_local_port_range > > echo 4096 > /proc/sys/net/ipv4/tcp_max_syn_backlog > > echo 1 > /proc/sys/net/ipv4/ip_forward > > > > But looks what happens when it starts: > > > > Fri Mar 8 09:10:10 2002 [0x400]report_limits(): RLIMIT_DATA: 4294967295 > > Fri Mar 8 09:10:10 2002 [0x400]report_limits(): RLIMIT_NOFILE: 8196 > > Fri Mar 8 09:10:10 2002 [0x400]report_limits(): RLIMIT_CORE: 4294967295 > > Fri Mar 8 09:10:10 2002 [0x400]report_limits(): RLIMIT_NPROC: 4294967295 > > > > Could RLIMIT_NOFILE be the problem? > > > > ----- Original Message ----- > > From: "Vladimir Ivaschenko" <[EMAIL PROTECTED]> > > To: <[EMAIL PROTECTED]> > > Sent: Friday, March 08, 2002 8:41 AM > > Subject: Re: [OOPS] increasing the number of threads on linux > > > > > > > >>Probably it is a problem not with threads, but with sockets. You need to > >>increase number of maximum sockets (__FD_SETSIZE) in glibc (types.h) and > >>in kernel. > >> > >>Edward Millington wrote: > >> > >> > >>>HI there! Does anyone knows how I can increase the number of thread > >>>linux can handle for oops? I find that linux could handle up to 950+ > >>>thread fairly well. At around 980 threads, oops stops working. Is > >>>there a way to solve this? With this big problem, I am thing of going > >>>back to squid. > >>>Thank you very much. > >>> > >>>Best regards, > >>> > >>>Edward Millington. BSc, Network+ > >>>Systems Administrator > >>>Cariaccess Communications Ltd. > >>>Palm Plaza > >>>Wildey > >>>St. Michael > >>>Barbados > >>>1-246-430-7435 > >>>Fax : 1-246-431-0170 > >>>[EMAIL PROTECTED] > >>>www.cariaccess.com > >>> > >>-- > >>Best Regards > >>Vladimir Ivaschenko > >>Certified Linux Engineer (RHCE) > -- > Joe Cooper <[EMAIL PROTECTED]> > http://www.swelltech.com > Web Caching Appliances and Support > > ===================================================================== > If you would like to unsubscribe from this list send message to > [EMAIL PROTECTED] with "unsubscribe oops-eng" in message body. > Archive is accessible on http://lists.paco.net/oops-eng/ > ===================================================================== If you would like to unsubscribe from this list send message to [EMAIL PROTECTED] with "unsubscribe oops-eng" in message body. Archive is accessible on http://lists.paco.net/oops-eng/
