From alberto at mellanox.co.il Mon Jan 21 15:51:53 2008 From: alberto at mellanox.co.il (Albert Ozilov) Date: Mon, 21 Jan 2008 17:51:53 +0200 Subject: I am having problems using an IPV6 address Message-ID: <6C2C79E72C305246B504CBA17B5500C903286EF5@mtlexch01.mtl.com> Hi, I am using polygraph version 2.8.1, and I am having problems using an IPV6 address with the simple.pg workload, does anyone have a clue? Here is the error lines from the server: gadgets.cc:282: (s0) Success gadgets.cc:282: (s0) Success 000.00| FDs: 1024 out of 1024 FDs can be used; safeguard limit: 983 000.00| resource usage: CPU Usage: 8msec sys + 348msec user = 356msec Page faults with physical i/o: 0 000.00| group-id: 1133687c.22106c7c:00000004 pid: 27772 000.00| current time: 1200927434.364076 or Mon, 21 Jan 2008 14:57:14 GMT 000.00| fyi: PGL configuration stored (383bytes) 000.00| fyi: no bench selected with use(); will not attempt to create agent addresses Server.cc:97: (s22) Invalid argument error: the server `S101' cannot listen on [fe80::202:c902:21:4cb9]:80; (s22) Invalid argument 000.00| got 0 xactions and 0 errors 000.00| shutdown reason: fatal error Best Regards, Alberto. -------------- next part -------------- An HTML attachment was scrubbed... URL: From rick.jones2 at hp.com Tue Jan 22 18:13:49 2008 From: rick.jones2 at hp.com (Rick Jones) Date: Tue, 22 Jan 2008 10:13:49 -0800 Subject: I am having problems using an IPV6 address In-Reply-To: <6C2C79E72C305246B504CBA17B5500C903286EF5@mtlexch01.mtl.com> References: <6C2C79E72C305246B504CBA17B5500C903286EF5@mtlexch01.mtl.com> Message-ID: <4796325D.40605@hp.com> Albert Ozilov wrote: > > Hi, > > I am using polygraph version 2.8.1, and I am having problems using an > IPV6 address with the simple.pg workload, does anyone have a clue? > > *Here is the error lines from the server:* > > gadgets.cc:282: (s0) Success > > gadgets.cc:282: (s0) Success > > 000.00| FDs: 1024 out of 1024 FDs can be used; safeguard limit: 983 > 000.00| resource usage: > > CPU Usage: 8msec sys + 348msec user = 356msec > > Page faults with physical i/o: 0 > > 000.00| group-id: 1133687c.22106c7c:00000004 pid: 27772 000.00| current > time: 1200927434.364076 or Mon, 21 Jan 2008 14:57:14 GMT 000.00| fyi: > PGL configuration stored (383bytes) 000.00| fyi: no bench selected with > use(); will not attempt to create agent addresses > > Server.cc:97: (s22) Invalid argument > > error: the server `S101' cannot listen on [fe80::202:c902:21:4cb9]:80; > (s22) Invalid argument > > 000.00| got 0 xactions and 0 errors > > 000.00| shutdown reason: fatal error Is this under Linux? Linux requires setting the scope/index field of the sockaddr_in6 structure so it can pick from the possibly more than one link-local (fe80::) addresses on the system. The linux getaddrinfo() and getnameinfo() calls support a syntax for IPv6 addresses which adds an interface name to allow it to find the correct index to put in the sockaddr_in6 structure: fe80::202:c902:21:4cb9% for example: fe80::202:c902:21:4cb%eth2 otherwise, you need to use something other than link-local IPv6 addresses. rick jones From alberto at mellanox.co.il Wed Jan 23 09:15:40 2008 From: alberto at mellanox.co.il (Albert Ozilov) Date: Wed, 23 Jan 2008 11:15:40 +0200 Subject: I am having problems using an IPV6 address In-Reply-To: <4796325D.40605@hp.com> References: <6C2C79E72C305246B504CBA17B5500C903286EF5@mtlexch01.mtl.com> <4796325D.40605@hp.com> Message-ID: <6C2C79E72C305246B504CBA17B5500C903287A35@mtlexch01.mtl.com> Hi Rick, Thanks you're trying to help here. I am using Linux SLES10, You are right about the address syntax I tried ping6 fe80::202:c902:21:4cb%eth0 and it works fine (as you can see below), however polysrv fails to run when I use this address syntax, it fail on the parsing. Here is the output of ping6: Linux:~ # ping6 fe80::219:b9ff:fee2:1f89%eth1 PING fe80::219:b9ff:fee2:1f89(fe80::219:b9ff:fee2:1f89) from fe80::219:b9ff:fee2:1f89 eth1: 56 data bytes 64 bytes from fe80::219:b9ff:fee2:1f89: icmp_seq=1 ttl=64 time=0.032 ms 64 bytes from fe80::219:b9ff:fee2:1f89: icmp_seq=2 ttl=64 time=0.009 ms 64 bytes from fe80::219:b9ff:fee2:1f89: icmp_seq=3 ttl=64 time=0.007 ms 64 bytes from fe80::219:b9ff:fee2:1f89: icmp_seq=4 ttl=64 time=0.008 ms --- fe80::219:b9ff:fee2:1f89 ping statistics --- 4 packets transmitted, 4 received, 0% packet loss, time 3000ms rtt min/avg/max/mdev = 0.007/0.014/0.032/0.010 ms This is the line I put into the conf file (Server S): addresses = ['[fe80::219:b9ff:fee2:1f89%eth1]:80' ]; // where these robot agents will be created Here is the output when I try to run polysrv: sw224:~ # /usr/local/polygraph/bin/polysrv --config /usr/local/polygraph/114-115-800conn.conf --verb_lvl 10 gadgets.cc:282: (s0) Success gadgets.cc:282: (s0) Success gadgets.cc:282: (s0) Success InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] PolyApp.cc:720: (s0) Success 000.00| Command: /usr/local/polygraph/bin/polysrv --config /usr/local/polygraph/114-115-800conn.conf --verb_lvl 10 000.00| Configuration: version: 2.8.1 host_type: x86_64-pc-unknown-linux-gnuoldld verb_lvl: 10 dump: err, dump_size: 1.000KB notify: doorman_listen_at: doorman_send_to: label: [none] fd_limit: 983 config: /usr/local/polygraph/114-115-800conn.conf cfg_dirs: console: - log: [none] log_buf_size: -1Bytes store_working_set: [none] load_working_set: [none] sample_log: [none] sample_log_buf_size:-1Bytes accept_foreign_msgs:off stats_cycle: 5.00sec file_scan: poll priority_sched: 5 fake_hosts: delete_old_addrs: yes idle_tout: 5.00min local_rng_seed: 1 global_rng_seed: 1 unique_world: on ign_urls: off 000.00| Server content distributions: Server S101: content planned% likely% error% mean_sz_bytes some-content 100.00 100.00 0.00 13326.41 expected average server-side cachability: 80.00% expected average server-side object size: 13326.41Bytes 000.00| Phases: phase pop_beg pop_end load_beg load_end rec_beg rec_end smsg_beg smsg_end goal flags meas 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 8.00hour 000.00| StatsSamples: static stats samples: 0 dynamic stats samples: 0 gadgets.cc:282: (s0) Success gadgets.cc:282: (s0) Success 000.00| FDs: 1024 out of 1024 FDs can be used; safeguard limit: 983 000.00| resource usage: CPU Usage: 24msec sys + 352msec user = 376msec Page faults with physical i/o: 0 000.00| group-id: 1135b675.000e5c52:00000004 pid: 23634 000.00| current time: 1201078467.916584 or Wed, 23 Jan 2008 08:54:27 GMT 000.00| fyi: PGL configuration stored (614bytes) 000.00| fyi: no bench selected with use(); will not attempt to create agent addresses InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] InAddress.cc:173: inet_pton() can't parse [fe80::219:b9ff:fee2:1f89%eth1] 000.00| created 0 agents total /usr/local/polygraph/bin/polysrv: no Server matches local interface addresses Server addresses:['[fe80::219:b9ff:fee2:1f89%eth1]:80'] local addresses: ['127.0.0.1', '10.4.3.224', '11.4.3.224', '12.4.3.224', '12.4.3.224', '[::1]', '[fe80::219:b9ff:fee2:1f89]', '[fe80::202:c902:21:975d]', '[fe80::202:c902:21:975e]'] 000.00| got 0 xactions and 0 errors 000.00| shutdown reason: fatal error Best Regards, Alberto. -----Original Message----- From: Rick Jones [mailto:rick.jones2 at hp.com] Sent: Tuesday, January 22, 2008 8:14 PM To: Albert Ozilov Cc: users at web-polygraph.org Subject: Re: I am having problems using an IPV6 address Albert Ozilov wrote: > > Hi, > > I am using polygraph version 2.8.1, and I am having problems using an > IPV6 address with the simple.pg workload, does anyone have a clue? > > *Here is the error lines from the server:* > > gadgets.cc:282: (s0) Success > > gadgets.cc:282: (s0) Success > > 000.00| FDs: 1024 out of 1024 FDs can be used; safeguard limit: 983 > 000.00| resource usage: > > CPU Usage: 8msec sys + 348msec user = 356msec > > Page faults with physical i/o: 0 > > 000.00| group-id: 1133687c.22106c7c:00000004 pid: 27772 000.00| > current > time: 1200927434.364076 or Mon, 21 Jan 2008 14:57:14 GMT 000.00| fyi: > PGL configuration stored (383bytes) 000.00| fyi: no bench selected > with use(); will not attempt to create agent addresses > > Server.cc:97: (s22) Invalid argument > > error: the server `S101' cannot listen on [fe80::202:c902:21:4cb9]:80; > (s22) Invalid argument > > 000.00| got 0 xactions and 0 errors > > 000.00| shutdown reason: fatal error Is this under Linux? Linux requires setting the scope/index field of the sockaddr_in6 structure so it can pick from the possibly more than one link-local (fe80::) addresses on the system. The linux getaddrinfo() and getnameinfo() calls support a syntax for IPv6 addresses which adds an interface name to allow it to find the correct index to put in the sockaddr_in6 structure: fe80::202:c902:21:4cb9% for example: fe80::202:c902:21:4cb%eth2 otherwise, you need to use something other than link-local IPv6 addresses. rick jones From rick.jones2 at hp.com Wed Jan 23 17:39:21 2008 From: rick.jones2 at hp.com (Rick Jones) Date: Wed, 23 Jan 2008 09:39:21 -0800 Subject: I am having problems using an IPV6 address In-Reply-To: <6C2C79E72C305246B504CBA17B5500C903287A35@mtlexch01.mtl.com> References: <6C2C79E72C305246B504CBA17B5500C903286EF5@mtlexch01.mtl.com> <4796325D.40605@hp.com> <6C2C79E72C305246B504CBA17B5500C903287A35@mtlexch01.mtl.com> Message-ID: <47977BC9.9070306@hp.com> Netperf doesn't use inet_pton, just getaddrinfo() so i've not seen that before. I'm not sure if that is a bug against inet_pton(), or an enhancement to be filed against polygraph. IMO if linux getaddrinfo() groks the %ethN then inet_pton() aught to at least be able to ignore the %ethN part. In the meantime though, it might be good to "wrap" inet_pton() in polygraph with a routing that ignores the %ethN part and calls the real inet_pton(). I've not looked at the source though - what does polygraph do with the results of inet_pton(). Depending on what it is trying to do, it may not be enough to ignore the %ethN bits... rick jones