From Bugzilla Helper: User-Agent: Mozilla/4.78 [en] (X11; U; Linux 2.4.16-0.9smp i686) Description of problem: I upgraded my workstation RHLinux system from 7.1 to 7.2. When I ping any IP addr or hostname on my internal network all is ok. When I ping any IP or hostname on my internal net or on the internet from my Linux iDSL firewall, all is ok. When I do IP or host pings from my 2 Windows computers all is ok. When I ping my iDSL router address, gateway address, etc. all is ok. BUT WHEN I PING HOSTNAMES ON THE INTERNET FROM MY NEWLY UPGRADED 7.2 RHLINUX WORKSTATION, I GET VERY HIGH % PACKET LOSES 50% AND UP. BUT PING THE IP ADDRESSES INSTEAD AND PING SHOWS NO PROBLEM. Seems like some kind of lookup issue local to my newly upgraded RHLinux 7.2. Version-Release number of selected component (if applicable): How reproducible: Always Steps to Reproduce: 1. Upgr 7.1 to 7.2 plus Errata updates for 7.2 2. Ping internet HOSTNAMES versus IP addresses to see packet loss difference. Actual Results: [phb@boaz phb]$ ping redhat.com PING redhat.com (216.148.218.197) from 192.168.100.1 : 56(84) bytes of data. 64 bytes from 216.148.218.197: icmp_seq=0 ttl=236 time=109.236 msec 64 bytes from 216.148.218.197: icmp_seq=1 ttl=236 time=110.064 msec 64 bytes from 216.148.218.197: icmp_seq=2 ttl=236 time=109.567 msec 64 bytes from 216.148.218.197: icmp_seq=3 ttl=236 time=110.819 msec 64 bytes from 216.148.218.197: icmp_seq=4 ttl=236 time=110.325 msec --- redhat.com ping statistics --- 14 packets transmitted, 6 packets received, 57% packet loss round-trip min/avg/max/mdev = 109.236/109.993/110.819/0.508 ms [phb@boaz phb]$ [phb@boaz phb]$ [phb@boaz phb]$ ping 216.148.218.197 PING 216.148.218.197 (216.148.218.197) from 192.168.100.1 : 56(84) bytes of data. 64 bytes from 216.148.218.197: icmp_seq=0 ttl=236 time=110.816 msec 64 bytes from 216.148.218.197: icmp_seq=1 ttl=236 time=125.453 msec 64 bytes from 216.148.218.197: icmp_seq=2 ttl=236 time=111.148 msec 64 bytes from 216.148.218.197: icmp_seq=3 ttl=236 time=113.947 msec 64 bytes from 216.148.218.197: icmp_seq=4 ttl=236 time=109.515 msec --- 216.148.218.197 ping statistics --- 5 packets transmitted, 5 packets received, 0% packet loss round-trip min/avg/max/mdev = 109.515/114.175/125.453/5.836 ms Expected Results: Pinging hostnames should not yield packet losses. Additional info: This behavior only started since 7.2 upgrade. Note this workstation is the only one having this issue.
This has nothing to with xinetd. If it was a software problem (and not just your network, which I think is the most likely problem), it would be a kernel problem. Reassigning.
Ted - it is NOT my network - if you read my overview, you would see that when I ping anything internal including my firewall network cards and my idsl router and gateway etc. IP or HOSTnames, there's NO PROBLEM. Also from two other computers on my network - WIndows PCs' - I can ping hostnames and IP addresses both internally on my network, and externally on the internet WITH NO PROBLEM. IT'S ONLY FROM MY RHL 7.2 WORKSTATION PINGING EXTERNAL INTERNET HOSTNAMES I AM GETTING HIGH PACKET LOSS - BUT PINGING THE IP ADDRESSES IS FINE. This problem appears to COME and GO - like right now it is fine, but I could find no valid reason for its occurance. Besides this only started to occur since the 7.1->7.2 upgrade, and under 7.1 I did not see this problem. I understand inetd is not the culprit, probably kernel. I am using k2.4.16-0.9 for maximum ext3 patching. I understand you should re-assign to kernel guys....fine. Thanks!
there's a quick test you could do to see if it might be the kernel or not. Can you add the hostname/ip pair you're testing with to the /etc/hosts file. The reason I ask this is because if this would fix it, it would be a DNS thing, and the kernel has nothing to do with DNS. If the problem remains, it MIGHT be kernel; however the difference between "ping by name" and "ping by number" is strange. Also, it appears that the sequence numbers of the pings are ok (eg sequential); this would mean there's "just" a delay, not a random loss. could you try using "ping -n hostname" (eg the -n) ?
Ok Arjanv... here's the test results for you.... ======================================================== PING vs. PING -n RESULTS (before I put hostname into /etc/hosts) ======================================================== PING RESULTS FOR HOSTNAME ON INTERNET [root@boaz root]# ping trimatrix.com PING trimatrix.com (192.41.49.147) from 192.168.100.1 : 56(84) bytes of data. 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=0 ttl=236 time=142.451 msec 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=1 ttl=236 time=141.966 msec 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=2 ttl=236 time=142.450 msec 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=3 ttl=236 time=144.212 msec --- trimatrix.com ping statistics --- 9 packets transmitted, 5 packets received, 44% packet loss round-trip min/avg/max/mdev = 141.966/142.758/144.212/0.802 ms [root@boaz root]# PING -n RESULTS FOR HOSTNAME ON INTERNET [root@boaz root]# ping -n trimatrix.com PING trimatrix.com (192.41.49.147) from 192.168.100.1 : 56(84) bytes of data. 64 bytes from 192.41.49.147: icmp_seq=0 ttl=236 time=145.598 msec 64 bytes from 192.41.49.147: icmp_seq=1 ttl=236 time=144.899 msec 64 bytes from 192.41.49.147: icmp_seq=2 ttl=236 time=142.163 msec 64 bytes from 192.41.49.147: icmp_seq=3 ttl=236 time=145.408 msec 64 bytes from 192.41.49.147: icmp_seq=4 ttl=236 time=144.600 msec --- trimatrix.com ping statistics --- 5 packets transmitted, 5 packets received, 0% packet loss round-trip min/avg/max/mdev = 142.163/144.533/145.598/1.305 ms ============================================================ NOW I PUT trimatrix.com and IP into /etc/hosts AND HERE'S RESULTS ============================================================ PING HOSTNAME RESULT [root@boaz etc]# ping trimatrix.com PING trimatrix.com (192.41.49.147) from 192.168.100.1 : 56(84) bytes of data. 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=0 ttl=236 time=142.219 msec 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=1 ttl=236 time=143.150 msec 64 bytes from trimatrix.com (192.41.49.147): icmp_seq=2 ttl=236 time=142.787 msec --- trimatrix.com ping statistics --- 3 packets transmitted, 3 packets received, 0% packet loss round-trip min/avg/max/mdev = 142.219/142.718/143.150/0.580 ms [root@boaz etc]# PING -n HOSTNAME RESULT [root@boaz etc]# ping -n trimatrix.com PING trimatrix.com (192.41.49.147) from 192.168.100.1 : 56(84) bytes of data. 64 bytes from 192.41.49.147: icmp_seq=0 ttl=236 time=144.348 msec 64 bytes from 192.41.49.147: icmp_seq=1 ttl=236 time=144.370 msec 64 bytes from 192.41.49.147: icmp_seq=2 ttl=236 time=147.068 msec 64 bytes from 192.41.49.147: icmp_seq=3 ttl=236 time=143.247 msec --- trimatrix.com ping statistics --- 4 packets transmitted, 4 packets received, 0% packet loss round-trip min/avg/max/mdev = 143.247/144.758/147.068/1.434 ms ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ALL THE ABOVE RESULTS ARE CONSISTANT. FURTHER, PINGING IP ADDRESS NEVER SHOWS PACKET LOSS. ONLY HOSTNAME WITHOUT -n OPTION WHEN NOT IN /etc/hosts, AND ONLY HOSTNAMES WHICH ARE ON THE INTERNET FROM THIS ONE RHL 7.2 WORKSTATION -ONLY-. BTW THERE'S NOTHING WRONG WITH trimatrix.com host ... I could have used any other internet host - same results.... ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Ok, this means it's a "ping" bug, not a kernel bug; changing to that component.....
OK, rawhide has an updated version of iptuils which contains a new version of ping. It changed quite a bit, so if you could give that version a try to see if it fixes your problem i'd greately appreciate it. I haven't been able to reproduce your problem btw. here, neither with the old nor with the new iputils package. Thanks, Read ya, Phil
As i can't reproduce this problem and haven't heard back i'll close this bug as WORKSFORME now. Read ya, Phil