| [Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] |
The problem with the above 2 is that Core-ISCSI is trying to send 32k PDUs while IET only accepts 8K PDUs. So I replaced it with the following: CHANNEL="0 1 lo 127.0.0.1 3260 0 AuthMethod=None;MaxRecvDataSegmentLength=8192 nopout_timeout=5" which should send 8K PDUs instead. In either case, I added a tcpdump of the login process at the end of my email. I then ran the same dd as before which should write 500 MB of sequential data onto the iSCSI target. It's talking a very long time to dd 500M onto the iSCSI target disc and looking at top, it gives me the following: istd1 taking 99.9% CPU with them split up as 50.1% sy and 49.9% wait Looking into /var/log/messages, I see messages such as: Mar 27 17:25:21 localhost kernel: iCHANNEL[0] - No defined iSCSI Authentication Methods, skipping SecurityNegotiation phase. Mar 27 17:25:46 localhost kernel: iscsi_login_rx_data:54: ***ERROR*** rx_data returned -512, expecting 48. Mar 27 17:25:46 localhost kernel: iscsi_initiator_start_negotiation:754: ***ERROR*** iSCSI Login negotiation failed. Mar 27 17:25:46 localhost kernel: iscsi_handle_authetication:108: ***ERROR*** sock_recvmsg() returned 0 Mar 27 17:25:48 localhost kernel: iCHANNEL[0] - No defined iSCSI Authentication Methods, skipping SecurityNegotiation phase. Mar 27 17:26:13 localhost kernel: iscsi_login_rx_data:54: ***ERROR*** rx_data returned -512, expecting 48. Mar 27 17:26:13 localhost kernel: iscsi_initiator_start_negotiation:754: ***ERROR*** iSCSI Login negotiation failed. Mar 27 17:26:13 localhost kernel: iscsi_handle_authetication:108: ***ERROR*** sock_recvmsg() returned 0 Mar 27 17:26:15 localhost kernel: iCHANNEL[0] - No defined iSCSI Authentication Methods, skipping SecurityNegotiation phase. ............. Could it be, as you mentioned in a previous mail, that since IET doesn't get back to the initiator and tell it that it's successfully written the data, the core-iscsi initiator believes that the connection is lost and tries to start a new session and fails because there's already the old session and IET doesn't support MCS? I'm off to try open-iscsi. Thanks for your help. Rik Herrin --- Ming Zhang <mingz@xxxxxxxxxxx> wrote: > On Sun, 2006-03-26 at 11:47 -0800, Rik Herrin wrote: > > > My /etc/sysconfig/initiator looks like this. > > > > CHANNEL="0 1 lo 127.0.0.1 3260 0" > > > > It would give me a lot of messages such as: > > > > check_segment_length(1552) too lond data 15a 32768 > > 8192 > > check_segment_length(1552) too lond data 15b 32768 > > 8192 > > looks like iet try to only accept pdu with 8K while > ini send out 32K. > could u send us a tcpdump capture of the login > process? > > we solve this first. > > > iscsi_session_push_cmnd(1531) too large cmd_sn > (56,1) > > ... [and so forth] > > > > before finally completing: > > > > real 1m31.320s > > user 0m0.001s > > sys 0m0.139s > > > > Sometimes it finishes while at other times, it > gives > > the errors above and then goes into a loop trying > to > > reconnect and then disconnecting again: > > > > localhost kernel: iCHANNEL[0]_LU[0] - SCSI Task > > Timeout with ITT: 0x00000264 > > localhost kernel: iCHANNEL[0]_LU[0] - Forcing > Session > > Reinstatement for timed-out SCSI Task to ONLINE > > Logical Unit > > localhost kernel: iCHANNEL[0] - Performing Cleanup > for > > failed iSCSI CID: 0 to > > iqn.2001-04.com.example:storage.disk2.sys1.xyz > > localhost kernel: iCHANNEL[0] - Decremented iSCSI > > connection count to 0 to node: > > iqn.2001-04.com.example:storage.disk2.sys1.xyz > > localhost kernel: iCHANNEL[0] - released iSCSI > session > > to node: > > iqn.2001-04.com.example:storage.disk2.sys1.xyz > > localhost kernel: iSCSI Core Stack[1] - > Decremented > > number of active iSCSI sessions to 0 > > localhost kernel: iCHANNEL[0] - No defined iSCSI > > Authentication Methods, skipping > SecurityNegotiation > > phase. > > localhost kernel: iscsi_login_rx_data:54: > ***ERROR*** > > rx_data returned -512, expecting 48. > > localhost kernel: > > iscsi_initiator_start_negotiation:754: ***ERROR*** > > iSCSI Login negotiation failed. > > localhost kernel: iscsi_handle_authetication:108: > > ***ERROR*** sock_recvmsg() returned 0 > > localhost kernel: iCHANNEL[0] - No defined iSCSI > > Authentication Methods, skipping > SecurityNegotiation > > phase. > > localhost kernel: iscsi_login_rx_data:54: > ***ERROR*** > > rx_data returned -512, expecting 48. > > localhost kernel: > > iscsi_initiator_start_negotiation:754: ***ERROR*** > > iSCSI Login negotiation failed. > > localhost kernel: iscsi_handle_authetication:108: > > ***ERROR*** sock_recvmsg() returned 0 > > localhost kernel: iCHANNEL[0] - No defined iSCSI > > Authentication Methods, skipping > SecurityNegotiation > > phase. > > localhost kernel: iscsi_login_rx_data:54: > ***ERROR*** > > rx_data returned -512, expecting 48. > > localhost kernel: > > iscsi_initiator_start_negotiation:754: ***ERROR*** > > iSCSI Login negotiation failed. > > localhost kernel: iscsi_handle_authetication:108: > > ***ERROR*** sock_recvmsg() returned 0 > > localhost kernel: iCHANNEL[0] - No defined iSCSI > > Authentication Methods, skipping > SecurityNegotiation > > phase. > > localhost kernel: iscsi_login_rx_data:54: > ***ERROR*** > > rx_data returned -512, expecting 48. > > localhost kernel: > > iscsi_initiator_start_negotiation:754: ***ERROR*** > > iSCSI Login negotiation failed. > > localhost kernel: iscsi_handle_authetication:108: > > ***ERROR*** sock_recvmsg() returned 0 > > localhost kernel: iCHANNEL[0] - No defined iSCSI > > Authentication Methods, skipping > SecurityNegotiation > > phase. > > localhost kernel: iscsi_login_rx_data:54: > ***ERROR*** > > rx_data returned -512, expecting 48. > > localhost kernel: > > iscsi_initiator_start_negotiation:754: ***ERROR*** > > iSCSI Login negotiation failed. > > localhost kernel: iscsi_handle_authetication:108: > > ***ERROR*** sock_recvmsg() returned 0 > > > > PS. I'll check out open-iscsi tomorrow morning and > > tell you how it goes. > > > > --- Ming Zhang <mingz@xxxxxxxxxxx> wrote: > > > > > forgot to say, the reason i ask u to check > > > open-iscsi is that we tested > > > open-iscsi + iet on same box before, it is not > ok, > > > then after a certain > > > version, open-iscsi can run ok with iet on same > box, > > > but nobody do > > > extensive tests. > > > > > > ming > > > > > > On Sun, 2006-03-26 at 12:37 -0500, Ming Zhang > wrote: > > > > frankly, i do not know. i assumed they should > be > > > ok no matter whether > > > > they are on same box or not. but from > experience, > > > it was not. > > > > > > > > maybe some kernel guru here can answer this > > > question. > > > > > > > > ming > > > > > > > > > > > > > > > > On Sun, 2006-03-26 at 09:18 -0800, Rik Herrin > > > wrote: > > > > > I'm running them on the same machine to see > the > > > > > maximum throughput and to see what type of > > > bottleneck > > > > > the disk subsystem will give me. I'm also > > > waiting for > > > > > a 4-port 1GbE Intel NIC to make sure that > the > > > network > > > > > doesn't end up being the bottleneck. Do you > > > think > > > > > there's any reason for them running on the > same > > > > > machine to cause a problem? > > > > > > > > > > Rik > > > > > > > > > > --- Ming Zhang <mingz@xxxxxxxxxxx> wrote: > > > > > > > > > > > ps, any point to run them at same box? ;) > > > > > > > > > > > > ming > > > > > > > > > > > > On Sun, 2006-03-26 at 08:30 -0800, Rik > Herrin > > > wrote: > > > > > > > I am going to try it with open-iscsi and > > > > > > core-iscsi to > > > > > > > see if this issue is initiator-specific > or > > > IET > > > > > > > specific. In the meantime, I configured > a > > > 4-disk > > > > > > RAID > > > > > > > 0 SATA configuration and set it as the > > > device > > > > > > exported > > > > > > > by IET. Without the iSCSI layer > (without > > > running > > > > > > > either ietd or iscsi-linux), I saw write > > > > > > throughput of > > > > > > > about 212 MB/s. When I started IET and > > > > > > linux-iscsi > > > > > > > and ran the sequential write test, I am > now > > > > > > getting > > > > > > > the following errors: > === message truncated === __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com ------------------------------------------------------- This SF.Net email is sponsored by xPML, a groundbreaking scripting language that extends applications into web and mobile media. Attend the live webcast and join the prime developer group breaking into this new coding territory! http://sel.as-us.falkag.net/sel?cmd=lnk&kid=110944&bid=241720&dat=121642 _______________________________________________ linux-iscsi-users mailing list linux-iscsi-users@xxxxxxxxxxxxxxxxxxxxx https://lists.sourceforge.net/lists/listinfo/linux-iscsi-users
[IP Storage] [IETF] [Linux SCSI] [iSCSI Book] [Linux Resources] [Yosemite News] [Photo] [Home] [IETF Announcements] [IETF Discussion] [SCSI Hardware]