EnigmA-X
2004-09-29, 04:05 AM CDT
Hi there,
I am running a Fedora Linux server at a colocation place in Rotterdam The Netherlands. The server runs Plesk 7 has a lot of options and possibilities and all is nice.
Until I found out a strange networking problem, first during FTP-transfers. When I downloaded a lot of small files (around 200 bytes), the connection was lost at 15 seconds + some random time. My FTP client was telling me there was a time-out.
I tried to figure out this problem, and while doing this I have seen that most of the ports used with the FTP transfer were staying in the CLOSE_WAIT state (I have monitored this with 'netstat -tc' and then started this FTP transfer).
First of all, I was thinking of a badly configured proftpd, but there was nothing to find and a couple of different configurations for proftpd didn't make the situation any better. At that point I found out the same problem exists with the Apache webserver: when many small files (like thumbnails) are loaded, no communication is possible anymore to my server. All ports must be in use, and those are staying in the CLOSE_WAIT state after a certain time-out (I guess). The problem is that's it very irritating (of course), but also, clients will get time-outs and during the time all ports are in the CLOSE_WAIT state I can't connect to my server in any way (not even SSH).
Does this problem sound familiar to anybody, does anyone have a good solution for this problem?
Some side info:
- Fedora Core release 1 (Yarrow)
- Linux srv00122.xxx.xxxx 2.4.22-1.2194.nptl #1 Thu Jun 17 10:58:40 EDT 2004 i686 i686 - i386 GNU/Linux
- Using xinetd
- :( No networking related hardware trouble:
eth0 Link encap:Ethernet HWaddr 00:0C:76:EF:8B:A3
inet addr:213.193.221.x Bcast:213.193.221.255 Mask:255.255.255.128
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:109467130 errors:0 dropped:10 overruns:0 frame:0
TX packets:2467787 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3266668050 (3115.3 Mb) TX bytes:1848311031 (1762.6 Mb)
Interrupt:11 Base address:0xe400
eth0:1 Link encap:Ethernet HWaddr 00:0C:76:EF:8B:A3
inet addr:213.193.221.x Bcast:213.193.221.255 Mask:255.255.255.128
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:109467130 errors:0 dropped:10 overruns:0 frame:0
TX packets:2467787 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3266668050 (3115.3 Mb) TX bytes:1848311031 (1762.6 Mb)
Interrupt:11 Base address:0xe400
I am running a Fedora Linux server at a colocation place in Rotterdam The Netherlands. The server runs Plesk 7 has a lot of options and possibilities and all is nice.
Until I found out a strange networking problem, first during FTP-transfers. When I downloaded a lot of small files (around 200 bytes), the connection was lost at 15 seconds + some random time. My FTP client was telling me there was a time-out.
I tried to figure out this problem, and while doing this I have seen that most of the ports used with the FTP transfer were staying in the CLOSE_WAIT state (I have monitored this with 'netstat -tc' and then started this FTP transfer).
First of all, I was thinking of a badly configured proftpd, but there was nothing to find and a couple of different configurations for proftpd didn't make the situation any better. At that point I found out the same problem exists with the Apache webserver: when many small files (like thumbnails) are loaded, no communication is possible anymore to my server. All ports must be in use, and those are staying in the CLOSE_WAIT state after a certain time-out (I guess). The problem is that's it very irritating (of course), but also, clients will get time-outs and during the time all ports are in the CLOSE_WAIT state I can't connect to my server in any way (not even SSH).
Does this problem sound familiar to anybody, does anyone have a good solution for this problem?
Some side info:
- Fedora Core release 1 (Yarrow)
- Linux srv00122.xxx.xxxx 2.4.22-1.2194.nptl #1 Thu Jun 17 10:58:40 EDT 2004 i686 i686 - i386 GNU/Linux
- Using xinetd
- :( No networking related hardware trouble:
eth0 Link encap:Ethernet HWaddr 00:0C:76:EF:8B:A3
inet addr:213.193.221.x Bcast:213.193.221.255 Mask:255.255.255.128
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:109467130 errors:0 dropped:10 overruns:0 frame:0
TX packets:2467787 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3266668050 (3115.3 Mb) TX bytes:1848311031 (1762.6 Mb)
Interrupt:11 Base address:0xe400
eth0:1 Link encap:Ethernet HWaddr 00:0C:76:EF:8B:A3
inet addr:213.193.221.x Bcast:213.193.221.255 Mask:255.255.255.128
UP BROADCAST RUNNING MULTICAST MTU:1500 Metric:1
RX packets:109467130 errors:0 dropped:10 overruns:0 frame:0
TX packets:2467787 errors:0 dropped:0 overruns:0 carrier:0
collisions:0 txqueuelen:1000
RX bytes:3266668050 (3115.3 Mb) TX bytes:1848311031 (1762.6 Mb)
Interrupt:11 Base address:0xe400