Re: Problems connecting to iSCSI target from bonded interface
On 12/07/2011 08:34 PM, Albert Chin wrote: On Wed, Dec 07, 2011 at 08:23:35PM -0600, Albert Chin wrote: On Wed, Dec 07, 2011 at 05:44:14PM -0600, Mike Christie wrote: Also for your kernel you will need to set the net.ipv4.conf.default.rp_filter to 0 or 2. What does this do? Found documentation for this but why is rp_filter=1 bad? That's what it's currently set to. Ditto for net.ipv4.conf.all.rp_filter. Are you recommending it because of the dropped packets we're seeing on bond1: Does it work with rp_filter=1 now again? In recent kernels with this commit: (git commit 27fed4175acf81ddd91d9a4ee2fd298981f60295 http://git.kernel.org/?p=linux/kernel/git/torvalds/linux-2.6.git;a=commit;h=27fed4175acf81ddd91d9a4ee2fd298981f60295) they changed the behavior so that you needed to set it to 0 or 2, or packets would not be delivered to the initiator when using iface binding which uses SO_BINDTODEVICE. -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: Problems connecting to iSCSI target from bonded interface
On Thu, Dec 08, 2011 at 02:37:48PM -0600, Mike Christie wrote: On 12/07/2011 08:34 PM, Albert Chin wrote: On Wed, Dec 07, 2011 at 08:23:35PM -0600, Albert Chin wrote: On Wed, Dec 07, 2011 at 05:44:14PM -0600, Mike Christie wrote: Also for your kernel you will need to set the net.ipv4.conf.default.rp_filter to 0 or 2. What does this do? Found documentation for this but why is rp_filter=1 bad? That's what it's currently set to. Ditto for net.ipv4.conf.all.rp_filter. Are you recommending it because of the dropped packets we're seeing on bond1: Does it work with rp_filter=1 now again? In recent kernels with this commit: Yes. As soon as I modified it to bind to bond1, it worke.d -- albert chin (ch...@thewrittenword.com) -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: Problems connecting to iSCSI target from bonded interface
On 12/06/2011 11:49 PM, Albert Chin wrote: # iscsiadm -m iface default tcp,empty,empty,empty,empty iser iser,empty,empty,empty,empty # iscsiadm -m iface -I iface0 --op=new New interface iface0 added # iscsiadm -m iface -I iface0 --op=update -n iface.hwaddress \ -v 00:1b:21:d3:f6:08 iface0 updated. # iscsiadm -m iface default tcp,empty,empty,empty,empty iser iser,empty,empty,empty,empty iface0 tcp,00:1b:21:d3:f6:08,empty,empty,empty Instead of hwaddress use the iface.net_ifacename setting. So delete the old iface iscsiadm -m iface -I iface -o delete then do iscsiadm -m iface -I iface0 --op=new iscsiadm -m iface -I iface0 --op=update -n iface.net_ifacename -v bond1 Then do discovery again then login. The iscsi tools will get confused and possibly pick the wrong interface to bind to, because bond1 has mac 00:1b:21:d3:f6:08 but then so did eth6 and we do not know which one you really meant. Also for your kernel you will need to set the net.ipv4.conf.default.rp_filter to 0 or 2. -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: Problems connecting to iSCSI target from bonded interface
On Wed, Dec 07, 2011 at 05:44:14PM -0600, Mike Christie wrote: On 12/06/2011 11:49 PM, Albert Chin wrote: # iscsiadm -m iface default tcp,empty,empty,empty,empty iser iser,empty,empty,empty,empty # iscsiadm -m iface -I iface0 --op=new New interface iface0 added # iscsiadm -m iface -I iface0 --op=update -n iface.hwaddress \ -v 00:1b:21:d3:f6:08 iface0 updated. # iscsiadm -m iface default tcp,empty,empty,empty,empty iser iser,empty,empty,empty,empty iface0 tcp,00:1b:21:d3:f6:08,empty,empty,empty Instead of hwaddress use the iface.net_ifacename setting. So delete the old iface iscsiadm -m iface -I iface -o delete then do iscsiadm -m iface -I iface0 --op=new iscsiadm -m iface -I iface0 --op=update -n iface.net_ifacename -v bond1 Then do discovery again then login. The iscsi tools will get confused and possibly pick the wrong interface to bind to, because bond1 has mac 00:1b:21:d3:f6:08 but then so did eth6 and we do not know which one you really meant. Yeah. Ran iscsid in debug mode and saw this. Thanks. Also for your kernel you will need to set the net.ipv4.conf.default.rp_filter to 0 or 2. What does this do? -- albert chin (ch...@thewrittenword.com) -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.
Re: Problems connecting to iSCSI target from bonded interface
On Wed, Dec 07, 2011 at 08:23:35PM -0600, Albert Chin wrote: On Wed, Dec 07, 2011 at 05:44:14PM -0600, Mike Christie wrote: Also for your kernel you will need to set the net.ipv4.conf.default.rp_filter to 0 or 2. What does this do? Found documentation for this but why is rp_filter=1 bad? That's what it's currently set to. Ditto for net.ipv4.conf.all.rp_filter. Are you recommending it because of the dropped packets we're seeing on bond1: # ifconfig bond1 bond1 Link encap:Ethernet HWaddr 00:1b:21:d3:f6:08 inet addr:10.191.58.9 Bcast:10.191.58.255 Mask:255.255.255.0 inet6 addr: fe80::21b:21ff:fed3:f608/64 Scope:Link UP BROADCAST RUNNING MASTER MULTICAST MTU:9000 Metric:1 RX packets:9839923 errors:0 dropped:4434 overruns:0 frame:0 TX packets:15240717 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:2224524704 (2.2 GB) TX bytes:81839661996 (81.8 GB) We see the same issue with dropped packets on the first bonded interface, bond0: # ifconfig bond0 bond0 Link encap:Ethernet HWaddr 00:1b:21:d3:f6:0a UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1 RX packets:15989558 errors:0 dropped:4436 overruns:0 frame:0 TX packets:8623601 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:15769948311 (15.7 GB) TX bytes:2909403062 (2.9 GB) -- albert chin (ch...@thewrittenword.com) -- You received this message because you are subscribed to the Google Groups open-iscsi group. To post to this group, send email to open-iscsi@googlegroups.com. To unsubscribe from this group, send email to open-iscsi+unsubscr...@googlegroups.com. For more options, visit this group at http://groups.google.com/group/open-iscsi?hl=en.