Hi. Can anybody explain why this is happening? Code: 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=1 ttl=243 time=150 ms 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=2 ttl=243 time=151 ms 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=3 ttl=243 time=151 ms 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=4 ttl=243 time=153 ms ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=7 ttl=243 time=153 ms 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=8 ttl=243 time=149 ms ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=11 ttl=243 time=152 ms 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=12 ttl=243 time=151 ms ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=16 ttl=243 time=152 ms ping: sendmsg: Operation not permitted 64 bytes from py-in-f99.google.com (64.233.167.99): icmp_seq=18 ttl=243 time=151 ms ping: sendmsg: Operation not permitted ping: sendmsg: Operation not permitted This happens for any site, not just google. The server runs mysql, php & apache and is quite busy. I think this may be causing connection errors for users also (including myself) so am quite keen to get this sorted, although the load is under 1 so I am suprised by the un-responsiveness of the pages served, and the pings going out! Thanks. edit: Ok so I stopped the httpd service and it pings ok now. So this error must occour becuase Apache is overloading the server? Can anybody give me any advice as how to stop this happening? Maybe increase the amount of active connections? Ta.
That is typically a networking problem, that is when the kernel refuses to transmit the packet, am suspecting congestion on the wire.
None, well apart from some 'file does not exist' errors caused by people not updating their bookmarks. Sounds fair, if this is the case what would my options be? And how can I test this is the case? edit: Do you mean my hosts bandwidth? Or my server? Thanks again for the help guys.
Might be a firewall problem. I've found some interesting links: http://mandrivausers.org/index.php?showtopic=18051 http://lists.danga.com/pipermail/memcached/2006-September/002726.html http://frankmash.blogspot.com/2005/12/ping-sendmsg-operation-not-permitted.html http://www.linuxquestions.org/quest.../ping-sendmsg-operation-not-permitted-307848/
One of the links referring to memchached I dont use (one of these days Il get my head round it and use it tho!) eaccelarator does me for now. Looking at this link: http://www.netfilter.org/documentation/FAQ/netfilter-faq-3.html#ss3.7 It says to increase the number of connectinos.... but my ip_conntrack_max is already set to 34576. Is it ok to up this? Cheers again
Tried disabling APF with no success, still the same issue. Stopped mysqld service (didnt think would make a difference but thought I better try) and again, no difference. Stopped httpd service, and voila, no more ping problems. So it def seems like its the number of connections coming from Apache that are causing the issue. Im going to look into apache optimization for now but if you have any quick tips I would apreciate it! Also see my top output Code: top - 14:06:23 up 38 days, 13:18, 1 user, load average: 1.48, 1.89, 1.46 Tasks: 154 total, 1 running, 153 sleeping, 0 stopped, 0 zombie Cpu(s): 31.9%us, 10.3%sy, 0.0%ni, 49.5%id, 3.0%wa, 0.0%hi, 5.3%si, 0.0%st Mem: 1994020k total, 1562692k used, 431328k free, 110104k buffers Swap: 2031608k total, 96k used, 2031512k free, 1040928k cached PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND 10388 mysql 15 0 563m 91m 4076 S 12.3 4.7 0:58.58 mysqld 10419 root 25 0 291m 9728 5288 S 1.0 0.5 0:00.58 httpd 11534 apache 15 0 292m 8756 3180 S 1.0 0.4 0:00.04 httpd 11543 apache 15 0 292m 8692 3180 S 1.0 0.4 0:00.04 httpd 11118 apache 15 0 294m 12m 5808 S 0.7 0.7 0:00.51 httpd 11126 apache 16 0 294m 13m 6332 S 0.7 0.7 0:00.56 httpd 11129 apache 15 0 292m 9024 3400 S 0.7 0.5 0:00.45 httpd 11172 apache 15 0 293m 11m 5416 S 0.7 0.6 0:00.24 httpd 11176 apache 15 0 293m 9140 3396 D 0.7 0.5 0:00.23 httpd 11185 apache 15 0 292m 9212 3528 S 0.7 0.5 0:00.25 httpd 11193 apache 15 0 294m 13m 6380 S 0.7 0.7 0:00.29 httpd 11197 apache 15 0 294m 12m 5716 S 0.7 0.7 0:00.29 httpd 11198 apache 15 0 294m 13m 6332 S 0.7 0.7 0:00.30 httpd 11217 apache 15 0 292m 8908 3260 S 0.7 0.4 0:00.24 httpd 11221 apache 15 0 292m 8904 3312 S 0.7 0.4 0:00.24 httpd 11527 apache 15 0 292m 8764 3180 S 0.7 0.4 0:00.06 httpd 11530 apache 15 0 292m 8736 3180 S 0.7 0.4 0:00.04 httpd 11533 apache 15 0 292m 8720 3172 S 0.7 0.4 0:00.05 httpd 11536 apache 15 0 292m 8712 3164 S 0.7 0.4 0:00.05 httpd 11538 apache 15 0 292m 8784 3164 S 0.7 0.4 0:00.03 httpd 11541 apache 15 0 292m 8720 3164 S 0.7 0.4 0:00.05 httpd 11544 apache 15 0 292m 8740 3188 S 0.7 0.4 0:00.02 httpd 11547 apache 16 0 292m 8708 3164 D 0.7 0.4 0:00.02 httpd 11548 apache 15 0 292m 8700 3164 S 0.7 0.4 0:00.02 httpd 11553 apache 15 0 292m 8664 3164 S 0.7 0.4 0:00.02 httpd 11092 root 15 0 12716 1104 788 R 0.3 0.1 0:00.54 top 11158 apache 15 0 294m 13m 6380 S 0.3 0.7 0:00.50 httpd 11186 apache 15 0 292m 8896 3232 S 0.3 0.4 0:00.25 httpd 11187 apache 15 0 292m 9488 3700 S 0.3 0.5 0:00.26 httpd 11205 apache 15 0 292m 8896 3220 S 0.3 0.4 0:00.23 httpd 11229 apache 15 0 294m 13m 5896 S 0.3 0.7 0:00.28 httpd 11528 apache 16 0 292m 8784 3196 S 0.3 0.4 0:00.03 httpd 11531 apache 15 0 292m 8712 3212 S 0.3 0.4 0:00.02 httpd 11535 apache 15 0 292m 8740 3176 S 0.3 0.4 0:00.04 httpd 11537 apache 15 0 294m 12m 5260 S 0.3 0.6 0:00.06 httpd 11540 apache 15 0 292m 8732 3180 S 0.3 0.4 0:00.03 httpd The number of httpd connections can vary dramatically from 1 to all that you can see above, which shoots up the amount of processor being used. Should this be happening?
Ok so I used this guide: http://www.howtoforge.com/configuring_apache_for_maximum_performance And it seems to have helped server speed but I am still getting the 'operation not permitted' errors when pinging out. I changed the following Code: KeepAlive Off KeepAliveTimeout 15 <IfModule prefork.c> StartServers 8 MinSpareServers 5 MaxSpareServers 20 ServerLimit 256 MaxClients 256 MaxRequestsPerChild 4000 </IfModule> to Code: KeepAlive On KeepAliveTimeout 5 <IfModule prefork.c> StartServers 20 MinSpareServers 30 MaxSpareServers 50 ServerLimit 256 MaxClients 256 MaxRequestsPerChild 3000 </IfModule> Do you think I should increase the ServerLimit & MaxClients variables too? Next step I guess is to work out what modules I need to keep enabled (all default modules are still being loaded). More reading for me I guess!
Tried that but still had the same problem. I also noticed increased load and memory usage and had to revert back to the original settings. Could there be some max connections limit in CENTOS? If found these links which although are way over the limit I need they might give me some idea. I will read through them and see if I can find any tips. http://www.stdlib.net/~colmmacc/Apachecon-EU2005/scaling-apache-presentation.pdf http://www.stdlib.net/~colmmacc/Apachecon-EU2005/scaling-apache-handout.pdf Cheers
