You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@httpd.apache.org by un...@riverstyx.net on 1999/04/08 23:20:46 UTC
apache-apr
Anyone know where I should be sending feedback about apache-apr?
current-testers doesn't seem to actually have any traffic. I've attempted
to get apache-apr running from a snapshot two days old, and while working
for a short period of time, it hangs.
---
tani hosokawa
river styx internet
Re: apache-apr
Posted by Ryan Bloom <rb...@raleigh.ibm.com>.
Sorry, I missed the zero. Set MaxRequestsPerChild to zero. That should
avoid the need to get out of fcntl. The fact that it takes longer to die
when MaxRequestsPerChild is higher, makes me think we are on the right
track.
Ryan
On Sun, 11 Apr 1999 unknown@riverstyx.net wrote:
> Well, I set it to 30, but the server hung after only about 2 minutes. If
> it helps at all, the connections are initially very fast (<1 sec), but
> then slowly take longer to connect, until eventually the browser will only
> get to the "contacted, waiting for reply". About 20 min. later, the
> server won't even accept connections.
>
> ---
> tani hosokawa
> river styx internet
>
>
> On Sun, 11 Apr 1999, Ryan Bloom wrote:
>
> >
> > Do me a favor, set MaxRequests per child. I saw an fcntl call, that was
> > most likely blocked. I think the child process was trying to quit,
but it
> > couldn't because it was stuck in fcntl. This is a known problem, and we
> > are currently discussing how to fix it.
> >
> > This basic problem, is that you cannot get a thread out of fcntl easily.
> > There are two options. pthread_cancel, and kill the whole process.
> > Because we are trying to kill the process when this problem occurs, the
> > second option isn't really all the bad. But, the second option requires
> > some busy-waiting to make sure all of our threads are in valid states, and
> > we don't forget about accepted connections.
> >
> > The pthread cancel call has other problems, that I will detail in a
> > message later, when I have more time to write the message.
> >
> > Ryan
> >
> > On Fri, 9 Apr 1999 unknown@riverstyx.net wrote:
> >
> > > On Fri, 9 Apr 1999, Manoj Kasichainula wrote:
> > >
> > > > On Thu, Apr 08, 1999 at 02:52:31PM -0700, unknown@riverstyx.net wrote:
> > > > > Well, I compiled it and got it going, had to make a couple small changes
> > > > > -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> > > > > are appropriate for a threaded Apache.
> > > >
> > > > They seem reasonable. There are lots of factors that go into good
> > > > selection of these numbers, and we still need to choose good defaults.
> > > >
> > > > > it didn't
> > > > > work all that great. Right now I've got it set up with 16 threads/child,
> > > > > maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> > > > > holding steady at 40-50 processes (sometimes spiking to 80). When I
> > > > > switch to apache-apr, it runs fine for a while, sometimes half an hour or
> > > > > so, but eventually it seems to hang. I connect, but don't actually manage
> > > > > to get a request through.
> > > >
> > > > The joys of threaded debugging, fun fun fun! I have lots and lots of
> > > > questions.
> > > >
> > > > Do you have any more details on the symptoms? Does the mod_status
> > > > display say anything interesting? Have you tried attaching a debugger
> > > > to one of the threads to see what's going on?
> > >
> > > Well, mod_status tells me that I'm throwing 200k/sec through it, but once
> > > the server hangs I can't get to mod_status anymore. I've tried a couple
> > > things, strace'ing the main server and the children but that doesn't seem
> > > to do much. Seems that the children just sit there, so I'm guessing
> > > strace doesn't work on threads. I've never had to debug a threaded app
> > > before, so I'm kinda in the dark here...
> > >
> > > > Is this with a "real load" or with benchmark tools? We've been
> > > > playing with benchmark tools like WebStone and ApacheBench, so a more
> > > > realistic load may expose problems we haven't hit yet.
> > >
> > > It's all real traffic.
> > >
> > > > What versions of glibc and the kernel are you running, and with what
> > > > Linux distribution? The threading support has been stabilizing
> > > > throughout the 2.0.x glibc series.
> > >
> > > The kernel's 2.2.5, RedHat 5.2 based, everything else is pretty much
> > > stock. I think glibc is 2.0.7.
> > >
> > > > Also, are you just pulling straight static content, or are you doing
> > > > intersting things like CGI or server-side includes? How about other
> > > > modules like mod_rewrite?
> > >
> > > I've got some CGI running, although not very often -- just a user
> > > interface for editing a web page. I'm using SSI to insert advertising
> > > headers into everyone's page. I'm using mod_rewrite to block image access
> > > from sites other than mine (to prevent hotlinking, basically). In fact,
> > > here's a stripped Configuration:
> > >
> > > EXTRA_CFLAGS=-DHTTPD_ROOT=\"/usr/local/apache-apr\"
> > > EXTRA_LDFLAGS=
> > > EXTRA_LIBS=
> > > EXTRA_INCLUDES=
> > > EXTRA_DEPS=
> > > Rule SHARED_CORE=default
> > > Rule SHARED_CHAIN=default
> > > Rule SOCKS4=no
> > > Rule SOCKS5=no
> > > Rule IRIXNIS=no
> > > Rule IRIXN32=yes
> > > Rule PARANOID=no
> > > Rule WANTHSREGEX=default
> > > AddModule modules/standard/mod_env.o
> > > AddModule modules/standard/mod_log_config.o
> > > AddModule modules/standard/mod_mime.o
> > > AddModule modules/standard/mod_negotiation.o
> > > AddModule modules/standard/mod_status.o
> > > AddModule modules/standard/mod_include.o
> > > AddModule modules/standard/mod_autoindex.o
> > > AddModule modules/standard/mod_dir.o
> > > AddModule modules/standard/mod_cgi.o
> > > AddModule modules/standard/mod_asis.o
> > > AddModule modules/standard/mod_imap.o
> > > AddModule modules/standard/mod_actions.o
> > > AddModule modules/standard/mod_alias.o
> > > AddModule modules/standard/mod_rewrite.o
> > > AddModule modules/standard/mod_access.o
> > > AddModule modules/standard/mod_auth.o
> > > AddModule modules/standard/mod_setenvif.o
> > >
> > > > Bill and I are seeing problems with the hybrid server when run with a
> > > > low MaxRequestsPerChild. What is yours set to?
> > >
> > > This should pretty much answer that and a few other future questions:
> > >
> > > Timeout 300
> > > KeepAlive On
> > > MaxKeepAliveRequests 100
> > > KeepAliveTimeout 15
> > > MinSpareServers 5
> > > MaxSpareServers 10
> > > StartServers 10
> > > ThreadsPerChild 16
> > > MaxClients 14
> > > MaxRequestsPerChild 1000
> > > ExtendedStatus On
> > >
> > >
> > > >
> > > > --
> > > > Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> > > > "I mean how can a punk be cyber?" - Bruce Sterling
> > > >
> > >
> > > ---
> > > tani hosokawa
> > > river styx internet
> > >
> > >
> >
> > _______________________________________________________________________
> > Ryan Bloom rbb@raleigh.ibm.com
> > 4205 S Miami Blvd
> > RTP, NC 27709 It's a beautiful sight to see good dancers
> > doing simple steps. It's a painful sight to
> > see beginners doing complicated patterns.
> >
>
>
_______________________________________________________________________
Ryan Bloom rbb@raleigh.ibm.com
4205 S Miami Blvd
RTP, NC 27709 It's a beautiful sight to see good dancers
doing simple steps. It's a painful sight to
see beginners doing complicated patterns.
Re: apache-apr
Posted by un...@riverstyx.net.
Well, I set it to 30, but the server hung after only about 2 minutes. If
it helps at all, the connections are initially very fast (<1 sec), but
then slowly take longer to connect, until eventually the browser will only
get to the "contacted, waiting for reply". About 20 min. later, the
server won't even accept connections.
---
tani hosokawa
river styx internet
On Sun, 11 Apr 1999, Ryan Bloom wrote:
>
> Do me a favor, set MaxRequests per child. I saw an fcntl call, that was
> most likely blocked. I think the child process was trying to quit, but it
> couldn't because it was stuck in fcntl. This is a known problem, and we
> are currently discussing how to fix it.
>
> This basic problem, is that you cannot get a thread out of fcntl easily.
> There are two options. pthread_cancel, and kill the whole process.
> Because we are trying to kill the process when this problem occurs, the
> second option isn't really all the bad. But, the second option requires
> some busy-waiting to make sure all of our threads are in valid states, and
> we don't forget about accepted connections.
>
> The pthread cancel call has other problems, that I will detail in a
> message later, when I have more time to write the message.
>
> Ryan
>
> On Fri, 9 Apr 1999 unknown@riverstyx.net wrote:
>
> > On Fri, 9 Apr 1999, Manoj Kasichainula wrote:
> >
> > > On Thu, Apr 08, 1999 at 02:52:31PM -0700, unknown@riverstyx.net wrote:
> > > > Well, I compiled it and got it going, had to make a couple small changes
> > > > -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> > > > are appropriate for a threaded Apache.
> > >
> > > They seem reasonable. There are lots of factors that go into good
> > > selection of these numbers, and we still need to choose good defaults.
> > >
> > > > it didn't
> > > > work all that great. Right now I've got it set up with 16 threads/child,
> > > > maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> > > > holding steady at 40-50 processes (sometimes spiking to 80). When I
> > > > switch to apache-apr, it runs fine for a while, sometimes half an hour or
> > > > so, but eventually it seems to hang. I connect, but don't actually manage
> > > > to get a request through.
> > >
> > > The joys of threaded debugging, fun fun fun! I have lots and lots of
> > > questions.
> > >
> > > Do you have any more details on the symptoms? Does the mod_status
> > > display say anything interesting? Have you tried attaching a debugger
> > > to one of the threads to see what's going on?
> >
> > Well, mod_status tells me that I'm throwing 200k/sec through it, but once
> > the server hangs I can't get to mod_status anymore. I've tried a couple
> > things, strace'ing the main server and the children but that doesn't seem
> > to do much. Seems that the children just sit there, so I'm guessing
> > strace doesn't work on threads. I've never had to debug a threaded app
> > before, so I'm kinda in the dark here...
> >
> > > Is this with a "real load" or with benchmark tools? We've been
> > > playing with benchmark tools like WebStone and ApacheBench, so a more
> > > realistic load may expose problems we haven't hit yet.
> >
> > It's all real traffic.
> >
> > > What versions of glibc and the kernel are you running, and with what
> > > Linux distribution? The threading support has been stabilizing
> > > throughout the 2.0.x glibc series.
> >
> > The kernel's 2.2.5, RedHat 5.2 based, everything else is pretty much
> > stock. I think glibc is 2.0.7.
> >
> > > Also, are you just pulling straight static content, or are you doing
> > > intersting things like CGI or server-side includes? How about other
> > > modules like mod_rewrite?
> >
> > I've got some CGI running, although not very often -- just a user
> > interface for editing a web page. I'm using SSI to insert advertising
> > headers into everyone's page. I'm using mod_rewrite to block image access
> > from sites other than mine (to prevent hotlinking, basically). In fact,
> > here's a stripped Configuration:
> >
> > EXTRA_CFLAGS=-DHTTPD_ROOT=\"/usr/local/apache-apr\"
> > EXTRA_LDFLAGS=
> > EXTRA_LIBS=
> > EXTRA_INCLUDES=
> > EXTRA_DEPS=
> > Rule SHARED_CORE=default
> > Rule SHARED_CHAIN=default
> > Rule SOCKS4=no
> > Rule SOCKS5=no
> > Rule IRIXNIS=no
> > Rule IRIXN32=yes
> > Rule PARANOID=no
> > Rule WANTHSREGEX=default
> > AddModule modules/standard/mod_env.o
> > AddModule modules/standard/mod_log_config.o
> > AddModule modules/standard/mod_mime.o
> > AddModule modules/standard/mod_negotiation.o
> > AddModule modules/standard/mod_status.o
> > AddModule modules/standard/mod_include.o
> > AddModule modules/standard/mod_autoindex.o
> > AddModule modules/standard/mod_dir.o
> > AddModule modules/standard/mod_cgi.o
> > AddModule modules/standard/mod_asis.o
> > AddModule modules/standard/mod_imap.o
> > AddModule modules/standard/mod_actions.o
> > AddModule modules/standard/mod_alias.o
> > AddModule modules/standard/mod_rewrite.o
> > AddModule modules/standard/mod_access.o
> > AddModule modules/standard/mod_auth.o
> > AddModule modules/standard/mod_setenvif.o
> >
> > > Bill and I are seeing problems with the hybrid server when run with a
> > > low MaxRequestsPerChild. What is yours set to?
> >
> > This should pretty much answer that and a few other future questions:
> >
> > Timeout 300
> > KeepAlive On
> > MaxKeepAliveRequests 100
> > KeepAliveTimeout 15
> > MinSpareServers 5
> > MaxSpareServers 10
> > StartServers 10
> > ThreadsPerChild 16
> > MaxClients 14
> > MaxRequestsPerChild 1000
> > ExtendedStatus On
> >
> >
> > >
> > > --
> > > Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> > > "I mean how can a punk be cyber?" - Bruce Sterling
> > >
> >
> > ---
> > tani hosokawa
> > river styx internet
> >
> >
>
> _______________________________________________________________________
> Ryan Bloom rbb@raleigh.ibm.com
> 4205 S Miami Blvd
> RTP, NC 27709 It's a beautiful sight to see good dancers
> doing simple steps. It's a painful sight to
> see beginners doing complicated patterns.
>
Re: apache-apr
Posted by Ryan Bloom <rb...@raleigh.ibm.com>.
Do me a favor, set MaxRequests per child. I saw an fcntl call, that was
most likely blocked. I think the child process was trying to quit, but it
couldn't because it was stuck in fcntl. This is a known problem, and we
are currently discussing how to fix it.
This basic problem, is that you cannot get a thread out of fcntl easily.
There are two options. pthread_cancel, and kill the whole process.
Because we are trying to kill the process when this problem occurs, the
second option isn't really all the bad. But, the second option requires
some busy-waiting to make sure all of our threads are in valid states, and
we don't forget about accepted connections.
The pthread cancel call has other problems, that I will detail in a
message later, when I have more time to write the message.
Ryan
On Fri, 9 Apr 1999 unknown@riverstyx.net wrote:
> On Fri, 9 Apr 1999, Manoj Kasichainula wrote:
>
> > On Thu, Apr 08, 1999 at 02:52:31PM -0700, unknown@riverstyx.net wrote:
> > > Well, I compiled it and got it going, had to make a couple small changes
> > > -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> > > are appropriate for a threaded Apache.
> >
> > They seem reasonable. There are lots of factors that go into good
> > selection of these numbers, and we still need to choose good defaults.
> >
> > > it didn't
> > > work all that great. Right now I've got it set up with 16 threads/child,
> > > maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> > > holding steady at 40-50 processes (sometimes spiking to 80). When I
> > > switch to apache-apr, it runs fine for a while, sometimes half an hour or
> > > so, but eventually it seems to hang. I connect, but don't actually manage
> > > to get a request through.
> >
> > The joys of threaded debugging, fun fun fun! I have lots and lots of
> > questions.
> >
> > Do you have any more details on the symptoms? Does the mod_status
> > display say anything interesting? Have you tried attaching a debugger
> > to one of the threads to see what's going on?
>
> Well, mod_status tells me that I'm throwing 200k/sec through it, but once
> the server hangs I can't get to mod_status anymore. I've tried a couple
> things, strace'ing the main server and the children but that doesn't seem
> to do much. Seems that the children just sit there, so I'm guessing
> strace doesn't work on threads. I've never had to debug a threaded app
> before, so I'm kinda in the dark here...
>
> > Is this with a "real load" or with benchmark tools? We've been
> > playing with benchmark tools like WebStone and ApacheBench, so a more
> > realistic load may expose problems we haven't hit yet.
>
> It's all real traffic.
>
> > What versions of glibc and the kernel are you running, and with what
> > Linux distribution? The threading support has been stabilizing
> > throughout the 2.0.x glibc series.
>
> The kernel's 2.2.5, RedHat 5.2 based, everything else is pretty much
> stock. I think glibc is 2.0.7.
>
> > Also, are you just pulling straight static content, or are you doing
> > intersting things like CGI or server-side includes? How about other
> > modules like mod_rewrite?
>
> I've got some CGI running, although not very often -- just a user
> interface for editing a web page. I'm using SSI to insert advertising
> headers into everyone's page. I'm using mod_rewrite to block image access
> from sites other than mine (to prevent hotlinking, basically). In fact,
> here's a stripped Configuration:
>
> EXTRA_CFLAGS=-DHTTPD_ROOT=\"/usr/local/apache-apr\"
> EXTRA_LDFLAGS=
> EXTRA_LIBS=
> EXTRA_INCLUDES=
> EXTRA_DEPS=
> Rule SHARED_CORE=default
> Rule SHARED_CHAIN=default
> Rule SOCKS4=no
> Rule SOCKS5=no
> Rule IRIXNIS=no
> Rule IRIXN32=yes
> Rule PARANOID=no
> Rule WANTHSREGEX=default
> AddModule modules/standard/mod_env.o
> AddModule modules/standard/mod_log_config.o
> AddModule modules/standard/mod_mime.o
> AddModule modules/standard/mod_negotiation.o
> AddModule modules/standard/mod_status.o
> AddModule modules/standard/mod_include.o
> AddModule modules/standard/mod_autoindex.o
> AddModule modules/standard/mod_dir.o
> AddModule modules/standard/mod_cgi.o
> AddModule modules/standard/mod_asis.o
> AddModule modules/standard/mod_imap.o
> AddModule modules/standard/mod_actions.o
> AddModule modules/standard/mod_alias.o
> AddModule modules/standard/mod_rewrite.o
> AddModule modules/standard/mod_access.o
> AddModule modules/standard/mod_auth.o
> AddModule modules/standard/mod_setenvif.o
>
> > Bill and I are seeing problems with the hybrid server when run with a
> > low MaxRequestsPerChild. What is yours set to?
>
> This should pretty much answer that and a few other future questions:
>
> Timeout 300
> KeepAlive On
> MaxKeepAliveRequests 100
> KeepAliveTimeout 15
> MinSpareServers 5
> MaxSpareServers 10
> StartServers 10
> ThreadsPerChild 16
> MaxClients 14
> MaxRequestsPerChild 1000
> ExtendedStatus On
>
>
> >
> > --
> > Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> > "I mean how can a punk be cyber?" - Bruce Sterling
> >
>
> ---
> tani hosokawa
> river styx internet
>
>
_______________________________________________________________________
Ryan Bloom rbb@raleigh.ibm.com
4205 S Miami Blvd
RTP, NC 27709 It's a beautiful sight to see good dancers
doing simple steps. It's a painful sight to
see beginners doing complicated patterns.
Re: apache-apr
Posted by un...@riverstyx.net.
On Fri, 9 Apr 1999, Manoj Kasichainula wrote:
> On Thu, Apr 08, 1999 at 02:52:31PM -0700, unknown@riverstyx.net wrote:
> > Well, I compiled it and got it going, had to make a couple small changes
> > -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> > are appropriate for a threaded Apache.
>
> They seem reasonable. There are lots of factors that go into good
> selection of these numbers, and we still need to choose good defaults.
>
> > it didn't
> > work all that great. Right now I've got it set up with 16 threads/child,
> > maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> > holding steady at 40-50 processes (sometimes spiking to 80). When I
> > switch to apache-apr, it runs fine for a while, sometimes half an hour or
> > so, but eventually it seems to hang. I connect, but don't actually manage
> > to get a request through.
>
> The joys of threaded debugging, fun fun fun! I have lots and lots of
> questions.
>
> Do you have any more details on the symptoms? Does the mod_status
> display say anything interesting? Have you tried attaching a debugger
> to one of the threads to see what's going on?
Well, mod_status tells me that I'm throwing 200k/sec through it, but once
the server hangs I can't get to mod_status anymore. I've tried a couple
things, strace'ing the main server and the children but that doesn't seem
to do much. Seems that the children just sit there, so I'm guessing
strace doesn't work on threads. I've never had to debug a threaded app
before, so I'm kinda in the dark here...
> Is this with a "real load" or with benchmark tools? We've been
> playing with benchmark tools like WebStone and ApacheBench, so a more
> realistic load may expose problems we haven't hit yet.
It's all real traffic.
> What versions of glibc and the kernel are you running, and with what
> Linux distribution? The threading support has been stabilizing
> throughout the 2.0.x glibc series.
The kernel's 2.2.5, RedHat 5.2 based, everything else is pretty much
stock. I think glibc is 2.0.7.
> Also, are you just pulling straight static content, or are you doing
> intersting things like CGI or server-side includes? How about other
> modules like mod_rewrite?
I've got some CGI running, although not very often -- just a user
interface for editing a web page. I'm using SSI to insert advertising
headers into everyone's page. I'm using mod_rewrite to block image access
from sites other than mine (to prevent hotlinking, basically). In fact,
here's a stripped Configuration:
EXTRA_CFLAGS=-DHTTPD_ROOT=\"/usr/local/apache-apr\"
EXTRA_LDFLAGS=
EXTRA_LIBS=
EXTRA_INCLUDES=
EXTRA_DEPS=
Rule SHARED_CORE=default
Rule SHARED_CHAIN=default
Rule SOCKS4=no
Rule SOCKS5=no
Rule IRIXNIS=no
Rule IRIXN32=yes
Rule PARANOID=no
Rule WANTHSREGEX=default
AddModule modules/standard/mod_env.o
AddModule modules/standard/mod_log_config.o
AddModule modules/standard/mod_mime.o
AddModule modules/standard/mod_negotiation.o
AddModule modules/standard/mod_status.o
AddModule modules/standard/mod_include.o
AddModule modules/standard/mod_autoindex.o
AddModule modules/standard/mod_dir.o
AddModule modules/standard/mod_cgi.o
AddModule modules/standard/mod_asis.o
AddModule modules/standard/mod_imap.o
AddModule modules/standard/mod_actions.o
AddModule modules/standard/mod_alias.o
AddModule modules/standard/mod_rewrite.o
AddModule modules/standard/mod_access.o
AddModule modules/standard/mod_auth.o
AddModule modules/standard/mod_setenvif.o
> Bill and I are seeing problems with the hybrid server when run with a
> low MaxRequestsPerChild. What is yours set to?
This should pretty much answer that and a few other future questions:
Timeout 300
KeepAlive On
MaxKeepAliveRequests 100
KeepAliveTimeout 15
MinSpareServers 5
MaxSpareServers 10
StartServers 10
ThreadsPerChild 16
MaxClients 14
MaxRequestsPerChild 1000
ExtendedStatus On
>
> --
> Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> "I mean how can a punk be cyber?" - Bruce Sterling
>
---
tani hosokawa
river styx internet
Re: apache-apr
Posted by Manoj Kasichainula <ma...@io.com>.
On Thu, Apr 08, 1999 at 02:52:31PM -0700, unknown@riverstyx.net wrote:
> Well, I compiled it and got it going, had to make a couple small changes
> -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> are appropriate for a threaded Apache.
They seem reasonable. There are lots of factors that go into good
selection of these numbers, and we still need to choose good defaults.
> it didn't
> work all that great. Right now I've got it set up with 16 threads/child,
> maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> holding steady at 40-50 processes (sometimes spiking to 80). When I
> switch to apache-apr, it runs fine for a while, sometimes half an hour or
> so, but eventually it seems to hang. I connect, but don't actually manage
> to get a request through.
The joys of threaded debugging, fun fun fun! I have lots and lots of
questions.
Do you have any more details on the symptoms? Does the mod_status
display say anything interesting? Have you tried attaching a debugger
to one of the threads to see what's going on?
Is this with a "real load" or with benchmark tools? We've been
playing with benchmark tools like WebStone and ApacheBench, so a more
realistic load may expose problems we haven't hit yet.
What versions of glibc and the kernel are you running, and with what
Linux distribution? The threading support has been stabilizing
throughout the 2.0.x glibc series.
Also, are you just pulling straight static content, or are you doing
intersting things like CGI or server-side includes? How about other
modules like mod_rewrite?
Bill and I are seeing problems with the hybrid server when run with a
low MaxRequestsPerChild. What is yours set to?
--
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
"I mean how can a punk be cyber?" - Bruce Sterling
Re: apache-apr
Posted by un...@riverstyx.net.
Oh, I know that... These aren't sites that I'm particularly concerned
about. I'm just interested in testing it out. I happen to have a nice
chunk of throw-away traffic that I can use it for, so.. here I am.
---
tani hosokawa
river styx internet
On Thu, 8 Apr 1999, Greg Stein wrote:
> Just in case it wasn't obvious... you *do* know that apache-apr
> shouldn't be used in a production environment, right?
>
> I believe testing has been done with a high transaction rate, but I
> don't know what sorts of persistent, high-load testing has been
> performed.
>
> Cheers,
> -g
>
> unknown@riverstyx.net wrote:
> >
> > Well, I compiled it and got it going, had to make a couple small changes
> > -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> > are appropriate for a threaded Apache. I'd been told that 32 threads was
> > normal for a Netscape webserver, so I tried it out... anyhow, it didn't
> > work all that great. Right now I've got it set up with 16 threads/child,
> > maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> > holding steady at 40-50 processes (sometimes spiking to 80). When I
> > switch to apache-apr, it runs fine for a while, sometimes half an hour or
> > so, but eventually it seems to hang. I connect, but don't actually manage
> > to get a request through. I was wondering how if it was just a matter of
> > tuning the server settings? Anyone else running with high load? I'm
> > using the Linux traffic shaper to hold my traffic down right now, but when
> > I open it up I immediately fill my 3 megabit connection to the top.
>
> --
> Greg Stein, http://www.lyra.org/
>
Re: apache-apr
Posted by Greg Stein <gs...@lyra.org>.
Just in case it wasn't obvious... you *do* know that apache-apr
shouldn't be used in a production environment, right?
I believe testing has been done with a high transaction rate, but I
don't know what sorts of persistent, high-load testing has been
performed.
Cheers,
-g
unknown@riverstyx.net wrote:
>
> Well, I compiled it and got it going, had to make a couple small changes
> -- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
> are appropriate for a threaded Apache. I'd been told that 32 threads was
> normal for a Netscape webserver, so I tried it out... anyhow, it didn't
> work all that great. Right now I've got it set up with 16 threads/child,
> maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
> holding steady at 40-50 processes (sometimes spiking to 80). When I
> switch to apache-apr, it runs fine for a while, sometimes half an hour or
> so, but eventually it seems to hang. I connect, but don't actually manage
> to get a request through. I was wondering how if it was just a matter of
> tuning the server settings? Anyone else running with high load? I'm
> using the Linux traffic shaper to hold my traffic down right now, but when
> I open it up I immediately fill my 3 megabit connection to the top.
--
Greg Stein, http://www.lyra.org/
Re: apache-apr
Posted by un...@riverstyx.net.
Well, I compiled it and got it going, had to make a couple small changes
-- HARD_SERVER_LIMIT = 30, HARD_THREAD_LIMIT = 64. I don't know if those
are appropriate for a threaded Apache. I'd been told that 32 threads was
normal for a Netscape webserver, so I tried it out... anyhow, it didn't
work all that great. Right now I've got it set up with 16 threads/child,
maxclients=14. I'm running the non-threaded right now (1.3.6) and it's
holding steady at 40-50 processes (sometimes spiking to 80). When I
switch to apache-apr, it runs fine for a while, sometimes half an hour or
so, but eventually it seems to hang. I connect, but don't actually manage
to get a request through. I was wondering how if it was just a matter of
tuning the server settings? Anyone else running with high load? I'm
using the Linux traffic shaper to hold my traffic down right now, but when
I open it up I immediately fill my 3 megabit connection to the top.
---
tani hosokawa
river styx internet
On Thu, 8 Apr 1999, Manoj Kasichainula wrote:
> On Thu, Apr 08, 1999 at 02:20:46PM -0700, unknown@riverstyx.net wrote:
> > Anyone know where I should be sending feedback about apache-apr?
> > current-testers doesn't seem to actually have any traffic. I've attempted
> > to get apache-apr running from a snapshot two days old, and while working
> > for a short period of time, it hangs.
>
> Feedback, patches, etc. can go to this list. Can you provide some
> details on your problems?
>
> --
> Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
> "This is not a novel to be tossed aside lightly. It should be thrown with
> great force." -- Dorothy Parker
>
Re: apache-apr
Posted by Manoj Kasichainula <ma...@io.com>.
On Thu, Apr 08, 1999 at 02:20:46PM -0700, unknown@riverstyx.net wrote:
> Anyone know where I should be sending feedback about apache-apr?
> current-testers doesn't seem to actually have any traffic. I've attempted
> to get apache-apr running from a snapshot two days old, and while working
> for a short period of time, it hangs.
Feedback, patches, etc. can go to this list. Can you provide some
details on your problems?
--
Manoj Kasichainula - manojk at io dot com - http://www.io.com/~manojk/
"This is not a novel to be tossed aside lightly. It should be thrown with
great force." -- Dorothy Parker