Re: Should connection restored?
HIMANSHU wrote: Yeah mike,This time you hit the Bulls eye. So only way(from initiators side) to successfully relogin to blocked target was increasing timeout value which i tried. Now we have to change the IET code to allow relogin.right? IET people are not really responding to my posts.Do you have any vague idea what things should be changed. I do not think IET needs any changes. If I restart IET, it just lets me log back in with no changes. Or if I do /etc/init.d/iscsi-target stop wait a long time /etc/init.d/iscsi-target start the initiator logs back in. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
HIMANSHU wrote: Yeah..I am using IET. I was getting session recovery timed out after 120 secs when it was 120. Now as it is 86400,I never observed after 86400sec. that session recovery timed out after 86400 secs. Otherwise is it NORMAL behavior of IET that the connection is lost of existing targets after restarting target daemon? I am not sure what you mean by connection is lost. If you mean that you see those 1011 errors then yes. When IET and open-iscsi are running normally we have a tcp connection to the IET. When you reboot IET, it closes it, so we try to reconnect. When we detect the problem we spit out an error 1011. If a nop/ping times out first though you might see a slightly different error, but normally when the tcp connection changes state we are notified and you see 1011 errors. If when you say the connection is lost is that we cannot do disk IO and you get FS errors and the iscsi session and its disks are basically dead and unusable because they only spit out errors then it is sort of expected :) We will not begin to fail IO until that replacement/recovery timer expires. So if you reboot IET and it takes longer than that timer to relogin then you will get FS/IO/SCSI/BLOCK errors. In Open-e,it was the case that disk blocked only some time,after it was recovered. But in my case,it is blocked forever. When you say blocked, do you mean if you run iscsiadm -m session -P 3 that the disks state says blocked or do you mean something else? Like I said before, when the connection is detected we block the scsi disks. When we are logged back or if the replacement/recovery timer fires we unblock them. If we are logged in we execte IO. If we are not then we fail IO. Open-e might changed IET to suit this persistent connection or they might be using something else?Thoughts - --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
The Rule: Only after logout,target daemon can be restarted. But it sounds little weird.is it? It looks like a safety feature. If some I/O going on the initiators side,i cannot logout that target. You sure about that? Did you do 'iscsiadm -m node -U all' and the session wouldn't logout? Without restarting target daemon,we can not make target/LUN addition,modification changed to reflect on initiators on discovery commmand. From the initiator side that is not true. The initiator can get new targets, rescan the LUNs, find new devices, etc. The limitation is in the iscsi-target software which doesn't allow you to dynamically do these things. If you have the experience you could write this functionality and propose it to the iscsi-target folks. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
Thanks a lot Konrad and Mike for your continuous help. @Konrad Sorry, I meant to say If I/O going on,it shouldn't be allowed to logout that target. Suppose I added few targets and want initiators should know this change,i should restart iscsi-target daemon.Without this,iscsiadm - m discovery. cannot detect addition of these new targets..It still shows OLD ones. Does it mean according to you,addition of new target should be spread by target itself or discovery command should sense it?? How initiator can sense addition of new targets other than discovery? Are there any special commands?I am using IET. LUN addition can be sensed by --rescan.But after restarting iscsi- target,as disks status is blocked,--rescan command hangs up completely. so after restarting iscsi-target,New target/LUN's can be sensed,But original connections are lost. Is it normal behavior of IET? -- @Mike Yeah..I am using IET. I was getting session recovery timed out after 120 secs when it was 120. Now as it is 86400,I never observed after 86400sec. that session recovery timed out after 86400 secs. Otherwise is it NORMAL behavior of IET that the connection is lost of existing targets after restarting target daemon? In Open-e,it was the case that disk blocked only some time,after it was recovered. But in my case,it is blocked forever. Open-e might changed IET to suit this persistent connection or they might be using something else?Thoughts - --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
I changed timeout as follows. 1. iscsiadm --mode discovery --type sendtargets --portal 192.168.7.174 2. iscsiadm -m node -T iqn.2008-05.com.abcde:Tar1 -p 192.168.7.174:3260 --login 3. iscsiadm -m node -T iqn.2008-05.com.qualexsystems:Tar1 -p 192.168.7.174:3260 -o update -n node.session.timeo.replacement_timeout -v 82400 (got succedded) 4. I mounted disk exposed and started some i/o on it. 5. Then i restarted iscsi-target daemon output of dmesg after during step 4 - __journal_remove_journal_head: freeing b_frozen_data scsi 1:0:0:1: rejecting I/O to dead device Buffer I/O error on device sdh1, logical block 529 lost page write due to I/O error on sdh1 scsi2 : iSCSI Initiator over TCP/IP Vendor: IET Model: VIRTUAL-DISK Rev: 0 Type: Direct-Access ANSI SCSI revision: 04 SCSI device sdg: 630784 512-byte hdwr sectors (323 MB) sdg: Write Protect is off sdg: Mode Sense: 77 00 00 08 SCSI device sdg: drive cache: write through SCSI device sdg: 630784 512-byte hdwr sectors (323 MB) sdg: Write Protect is off sdg: Mode Sense: 77 00 00 08 SCSI device sdg: drive cache: write through sdg: sdg1 sd 2:0:0:0: Attached scsi disk sdg sd 2:0:0:0: Attached scsi generic sg6 type 0 Vendor: IET Model: VIRTUAL-DISK Rev: 0 Type: Direct-Access ANSI SCSI revision: 04 SCSI device sdh: 630784 512-byte hdwr sectors (323 MB) sdh: Write Protect is off sdh: Mode Sense: 77 00 00 08 SCSI device sdh: drive cache: write through SCSI device sdh: 630784 512-byte hdwr sectors (323 MB) sdh: Write Protect is off sdh: Mode Sense: 77 00 00 08 SCSI device sdh: drive cache: write through sdh: sdh1 sd 2:0:0:1: Attached scsi disk sdh sd 2:0:0:1: Attached scsi generic sg7 type 0 kjournald starting. Commit interval 5 seconds EXT3 FS on sdh1, internal journal EXT3-fs: mounted filesystem with ordered data mode. dmesg after step 5 session1: iscsi: session recovery timed out after 120 secs iscsi: cmd 0x2a is not queued (7) sd 2:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdh, sector 390208 Buffer I/O error on device sdh1, logical block 195073 lost page write due to I/O error on sdh1 Buffer I/O error on device sdh1, logical block 195074 lost page write due to I/O error on sdh1 Aborting journal on device sdh1. iscsi: cmd 0x2a is not queued (7) sd 2:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdh, sector 390208 Buffer I/O error on device sdh1, logical block 195073 lost page write due to I/O error on sdh1 Buffer I/O error on device sdh1, logical block 195074 lost page write due to I/O error on sdh1 __journal_remove_journal_head: freeing b_committed_data ext3_abort called. EXT3-fs error (device sdh1): ext3_journal_start_sb: Detected aborted journal Remounting filesystem read-only iscsi: cmd 0x2a is not queued (7) sd 2:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdh, sector 66 Buffer I/O error on device sdh1, logical block 2 lost page write due to I/O error on sdh1 iscsi: cmd 0x2a is not queued (7) iscsi: cmd 0x2a is not queued (7) sd 2:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdh, sector 376896 Buffer I/O error on device sdh1, logical block 188417 lost page write due to I/O error on sdh1 sd 2:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdh, sector 376900 Buffer I/O error on device sdh1, logical block 188419 lost page write due to I/O error on sdh1 --- What is exactly happening here? timeout value is changing from 120 to 86400..How can i re-confirm that? Can there be some other problem as well? - Target: iqn.2008-05.com.qualexsystems:Tar2 Current Portal: 192.168.7.174:3260,1 Persistent Portal: 192.168.7.174:3260,1 ** Interface: ** Iface Name: default Iface Transport: tcp Iface IPaddress: default Iface HWaddress: default Iface Netdev: default SID: 4 iSCSI Connection State: LOGGED IN Internal iscsid Session State: NO CHANGE Negotiated iSCSI params: HeaderDigest: CRC32C DataDigest: None MaxRecvDataSegmentLength: 131072 MaxXmitDataSegmentLength: 8192 FirstBurstLength: 65536 MaxBurstLength: 262144 ImmediateData: No InitialR2T: Yes MaxOutstandingR2T: 1
Re: Should connection restored?
HIMANSHU wrote: Hi.. It is probably not the problem of replacement_timer. Did you try what I asked? Did you do the same timing in the tests? In your log you had this: session13: iscsi: session recovery timed out after 120 secs When this is is printed out it means the replacment timer has fired and the devices will be unblocked (while they are blocked IO will be queued and will process will look like they are hung waiting for it) and IO will be failed until the session comes back. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
I) How to increase replacement_timeout on initiator? echo 82400 /sys/block/sdh/device/timeout or After discovery,nodes sendtargets are created.In nodes,there is default file. node.session.timeo.replacement_timeout = 120 node.session.err_timeo.abort_timeout = 10 node.session.err_timeo.reset_timeout = 30 node.session.iscsi.FastAbort = Yes node.conn[0].timeo.logout_timeout = 15 node.conn[0].timeo.login_timeout = 15 node.conn[0].timeo.auth_timeout = 45 node.conn[0].timeo.active_timeout = 5 node.conn[0].timeo.idle_timeout = 60 node.conn[0].timeo.ping_timeout = 5 node.conn[0].timeo.noop_out_interval = 10 node.conn[0].timeo.noop_out_timeout = 15 ... ... Can these values be useful to me?Or only replacement timeout HIGH value and node.conn[0].timeo.noop_out_interval = 0 node.conn[0].timeo.noop_out_timeout = 0 as given in documentation will do? And 1 more replacement timeout is not available in target side? Right.Only to be changed on the initiator side. Here's target's ietd file Target iqn.2008-04.com.qualexsystems:Tar1 Alias Tar1 Lun 1 Path=/dev/Vg1/Lv2,Type=fileio Lun 0 Path=/dev/Vg1/Lv1,Type=fileio HeaderDigest CRC32C,None InitialR2T Yes MaxBurstLength 262144 MaxRecvDataSegmentLength 8192 DataPDUInOrder Yes ImmediateData No MaxXmitDataSegmentLength 8192 FirstBurstLength 65536 MaxOutstandingR2T 8 DataSequenceInOrder Yes DataDigest CRC32C,None DefaultTime2Wait 2 MaxConnections 1 DefaultTime2Retain 20 ErrorRecoveryLevel 0 Wthreads 8 - II) After particular target iqn.2008-04.com.qualexsystems:Tar1 is logged in to an initiator,is it Ethical to add LUN's on that target? -- III) If CHAP Bidirectional authentication is is given to 2 targets Tar1 Tar2,Can same initiator machine can login to both targets? Because for 2nd targets login,/etc/iscsi/iscsid.conf on the initiator should be changed so that login can be possible.But here we are losing our previous target's authentication,when we overwrite iscsid.conf with Tar2s Uname Pwd. And still Tar1 is logged in though iscsid.conf doesn't contain it's Authentication parameters,but contain Tar2's.Is it accepted behavior or it is weird?Your views -- IV) We are not using Multipath.So i can't try your previous suggestion. Spring for an extra nic on the target and use multipath and set queue_if_no_path or no_path_retry queue. -- Is there any other way to make connections persistent?When i used open-e(DSS) as target machine,Initiator retained persistent connections after target restart.Our target restart code is already mentioned earlier. Thank you very much - --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
HIMANSHU wrote: I) How to increase replacement_timeout on initiator? echo 82400 /sys/block/sdh/device/timeout or Yeah, the timeout above is not the replacement_timeout so writing to that sysfs file will not help. After discovery,nodes sendtargets are created.In nodes,there is default file. node.session.timeo.replacement_timeout = 120 node.session.err_timeo.abort_timeout = 10 node.session.err_timeo.reset_timeout = 30 node.session.iscsi.FastAbort = Yes node.conn[0].timeo.logout_timeout = 15 node.conn[0].timeo.login_timeout = 15 node.conn[0].timeo.auth_timeout = 45 node.conn[0].timeo.active_timeout = 5 node.conn[0].timeo.idle_timeout = 60 node.conn[0].timeo.ping_timeout = 5 node.conn[0].timeo.noop_out_interval = 10 node.conn[0].timeo.noop_out_timeout = 15 ... ... Can these values be useful to me?Or only replacement timeout HIGH value and Yes. This it the replacement_timeout you want to set. Run iscsiadm -m node -T target -p ip:port -o update -n node.session.timeo.replacement_timeout -v 82400 You can also edit the file by hand, but I would use iscsiadm in general because it will handle changes in the file format for you. node.conn[0].timeo.noop_out_interval = 0 node.conn[0].timeo.noop_out_timeout = 0 as given in documentation will do? And 1 more replacement timeout is not available in target side? Right.Only to be changed on the initiator side. Yeah, right. There is not target setting. Here's target's ietd file Target iqn.2008-04.com.qualexsystems:Tar1 Alias Tar1 Lun 1 Path=/dev/Vg1/Lv2,Type=fileio Lun 0 Path=/dev/Vg1/Lv1,Type=fileio HeaderDigest CRC32C,None InitialR2T Yes MaxBurstLength 262144 MaxRecvDataSegmentLength 8192 DataPDUInOrder Yes ImmediateData No MaxXmitDataSegmentLength 8192 FirstBurstLength 65536 MaxOutstandingR2T 8 DataSequenceInOrder Yes DataDigest CRC32C,None DefaultTime2Wait 2 MaxConnections 1 DefaultTime2Retain 20 ErrorRecoveryLevel 0 Wthreads 8 - II) After particular target iqn.2008-04.com.qualexsystems:Tar1 is logged in to an initiator,is it Ethical to add LUN's on that target? Yes. To find them on the initiator side you then would need to do iscsiadm -m session --rescan -- III) If CHAP Bidirectional authentication is is given to 2 targets Tar1 Tar2,Can same initiator machine can login to both targets? Yes. Because for 2nd targets login,/etc/iscsi/iscsid.conf on the initiator should be changed so that login can be possible.But here we are losing our previous target's authentication,when we overwrite iscsid.conf with Tar2s Uname Pwd. And still Tar1 is logged in though iscsid.conf doesn't contain it's Authentication parameters,but contain Tar2's.Is it accepted behavior or it is weird?Your views iscsi.conf is only read when you do iscsiadm -m discovery . iscsiadm will read iscsi.conf and put those values in /etc/iscsi/nodes/target/portal/default. So if you had different CHAP values for the targets they would get their own default file in /etc/iscsi/nodes. -- IV) We are not using Multipath.So i can't try your previous suggestion. Spring for an extra nic on the target and use multipath and set queue_if_no_path or no_path_retry queue. -- Is there any other way to make connections persistent?When i used open-e(DSS) as target machine,Initiator retained persistent connections after target restart.Our target restart code is already mentioned earlier. Set the replacement_timeout how I describeed above. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
Hello All, 1. I am using latest iscsitarget-0.4.16-1 from sourceforge. Now,can you tell whether it is expected to retain the connection with initiator once the target is restarted? - 2. You wrote. The reason for having you do iscsid -d 8 was so we could see why we cannot log back in iscsid -d 8 log messages can be seen from dmesg...right? check 2nd last line here in dmesg.connection13:0: iscsi: detected conn error (1011) --- iscsi: cmd 0x28 is not queued (7) iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 0 sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 8 iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 0 iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 835576 iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 835576 iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 0 iscsi: cmd 0x28 is not queued (7) iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 8 iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 16 sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 0 iscsi: cmd 0x28 is not queued (7) sd 14:0:0:1: SCSI error: return code = 0x0001 end_request: I/O error, dev sdi, sector 0 scsi14 : iSCSI Initiator over TCP/IP Vendor: IET Model: VIRTUAL-DISK Rev: 0 Type: Direct-Access ANSI SCSI revision: 04 SCSI device sdh: 1048576 512-byte hdwr sectors (537 MB) sdh: Write Protect is off sdh: Mode Sense: 77 00 00 08 SCSI device sdh: drive cache: write through SCSI device sdh: 1048576 512-byte hdwr sectors (537 MB) sdh: Write Protect is off sdh: Mode Sense: 77 00 00 08 SCSI device sdh: drive cache: write through sdh: sdh1 sd 14:0:0:0: Attached scsi disk sdh sd 14:0:0:0: Attached scsi generic sg6 type 0 Vendor: IET Model: VIRTUAL-DISK Rev: 0 Type: Direct-Access ANSI SCSI revision: 04 SCSI device sdi: 835584 512-byte hdwr sectors (428 MB) sdi: Write Protect is off sdi: Mode Sense: 77 00 00 08 SCSI device sdi: drive cache: write through SCSI device sdi: 835584 512-byte hdwr sectors (428 MB) sdi: Write Protect is off sdi: Mode Sense: 77 00 00 08 SCSI device sdi: drive cache: write through sdi: sdi1 sd 14:0:0:1: Attached scsi disk sdi sd 14:0:0:1: Attached scsi generic sg7 type 0 connection13:0: iscsi: detected conn error (1011) session13: iscsi: session recovery timed out after 120 secs --- 3. when session is recovered. [EMAIL PROTECTED] iscsi]# iscsiadm -m session -P 3 iSCSI Transport Class version 1.1-646 iscsiadm version 2.0-865 Target: iqn.2008-04.com.qualexsystems:Tar1 Current Portal: 192.168.7.173:3260,1 Persistent Portal: 192.168.7.173:3260,1 ** Interface: ** Iface Name: default Iface Transport: tcp Iface IPaddress: default Iface HWaddress: default Iface Netdev: default SID: 13 iSCSI Connection State: IN LOGIN Internal iscsid Session State: REPOEN Negotiated iSCSI params: HeaderDigest: None DataDigest: None MaxRecvDataSegmentLength: 8192 MaxXmitDataSegmentLength: 8192 FirstBurstLength: 65536 MaxBurstLength: 262144 ImmediateData: No InitialR2T: Yes MaxOutstandingR2T: 1 Attached SCSI devices: Host Number: 14 State: running scsi14 Channel 00 Id 0 Lun: 0 Attached scsi disk sdh State: running scsi14 Channel 00 Id 0 Lun: 1 Attached scsi disk sdi State: running Though Disk state is running,still iSCSI Connection State: IN LOGIN. So disks still becomes unusable. Conclusion: What is moral of the story i
Re: Should connection restored?
On Fri, May 09, 2008 at 02:36:25AM -0700, MAKHU wrote: Hello All, 1. I am using latest iscsitarget-0.4.16-1 from sourceforge. And an older version of Open-iSCSI..would say 868-20. Have you tried using the one that got released about a week ago? ... snip ... Host Number: 14 State: running scsi14 Channel 00 Id 0 Lun: 0 Attached scsi disk sdh State: running scsi14 Channel 00 Id 0 Lun: 1 Attached scsi disk sdi State: running Though Disk state is running,still iSCSI Connection State: IN LOGIN. So disks still becomes unusable. Are the disks really unusable? What happens when you do 'dd if=/dev/sdh of=/dev/null count=1' ? --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
MAKHU wrote: Conclusion: What is moral of the story i got is that in our target- initiator pair,Persistent connection is not possible.After target/ Initiator iscsi-target daemon restart(code given below),connection is lost causing disks to be unusable. What you should have taken from the mails was that you should either: 1. Spring for an extra nic on the target and use multipath and set queue_if_no_path or no_path_retry queue. 2. set the iscsi node.session.timeo.replacement_timeout timer very high. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Should connection restored?
When target is logged in to an initiator and then either target/ initiator is restarted,connection is lost. Should the connection be restored? --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
On Thu, May 08, 2008 at 08:52:35AM -0700, MAKHU wrote: When target is logged in to an initiator and then either target/ initiator is restarted,connection is lost. It goes the other way. Initiator logs in the target. If the initiator (client) is restarted the connection would be lost. If the target is restarted it might do: 1). If it a NetApp, send a command asking the initiator to logout. 2). If is a EqualLogic, send a AsyncMsg telling the initiator that the block device is going to be off-line. 3). For others it might just terminate the connection without notifying the initiator at all. At which point the nop-ping timer (which runs by default every 15 seconds) would figure out the connection is lost, kick a retry and if that failed terminate the connection. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---
Re: Should connection restored?
Konrad Rzeszutek wrote: 2). If is a EqualLogic, send a AsyncMsg telling the initiator that the block device is going to be off-line. What is this? I do not think we handle this. Is it a verndor specific async iscsi event? Yes. To be exact attached is a TCP dump, look for seq no 1774. We do try to re-login afterwards. Ah that is just target requests logout. That is standard and like you said we handle it by trying to relogin. You worried me :) I just fixed that code in 754 or something and I thought I goofed and now they did an actual vendor specific one and we had to add some more new code to handle it. 3). For others it might just terminate the connection without notifying the initiator at all. At which point the nop-ping timer (which runs by default every 15 seconds) would figure out the connection is lost, kick a retry and if that failed terminate the connection. For 1 and 3, it depends on the version of open-iscsi you are using and what the target returns on the retry if it is able to send something at all. With the current open-iscsi code, if the tcp/ip socket connect fails we continue to retry that forever or until the user manually kills it. If when we try the relogin if the target is still up and responding and responds with target not found then we will kill the connection. If we get some other errors we will continue to retry the login. I thought the re-login routine did some back-off. Like 1 second, 2 seconds, then 4 seconds and so on.. Granted the attached dump shows it to try to No we use the def time2wait we got during login negotiation. We also are a little broken and we use the time2wait returned from a logout response when we are not supposed to. re-login non-stop and maybe I am confusing it with the 2.0-868-20 which had a nop-ping code in the iSCSI daemon instead in the kernel. The nop ping code does not change the relogin behavior here. You are probably remembering linux-iscsi. It will nicely back off. --~--~-~--~~~---~--~~ 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 [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/open-iscsi -~--~~~~--~~--~--~---