iSCSI storage domain and multipath when adding node

Hello, my configuration is what described here: http://lists.ovirt.org/pipermail/users/2017-March/080992.html So I'm using iSCSI multipath and not bonding can anyone reproduce? Initial situation is only one node configured and active with some VMS I go and configure a second node; it tries to activate but networks are not all already mapped and so gies to non operational. I setup all networks and activate the node It happens that: - on the first node where I currently have 2 iSCSI connections and 2multipath lines (with p1p1.100 and p1p2) it is instantiated a new iSCSI SID using interface "default" and in multipath -l output I see now 3 lines - on the newly added node I only see 1 iSCSI SID using interface default My way to solve the situation was to go inside iscsi multipath section do nothing but save the same config brutally on first node iscsiadm -m session -u --> all iscsi sessions are closed after a while I see again the original 2 connections recovered, with correct interface names used - on second node iscsiadm -m session -u --> the only session is cloed nothing happens if I set to maintenance the node and then activate the node --> the 2 correct iscsi sessions are activated... Thanks Gianluca

Gianluca, I can see that the workaround you describe here (to complete multipath configuration in CLI) fixes an inconsistency in observed iSCSI sessions. I think it is a shortcoming in oVirt that you had to resort to manual configuration. Could you file a bug about this? Ideally, following the bug template presented to you by Bugzilla, i.e. "Expected: two iSCSI sessions", "Got: one the first node ... one the second node". Edy, Martin, do you think you could help out here? Thanks, Ondra On Fri, Apr 7, 2017 at 5:21 PM, Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
Hello, my configuration is what described here: http://lists.ovirt.org/pipermail/users/2017-March/080992.html
So I'm using iSCSI multipath and not bonding can anyone reproduce?
Initial situation is only one node configured and active with some VMS
I go and configure a second node; it tries to activate but networks are not all already mapped and so gies to non operational. I setup all networks and activate the node
It happens that: - on the first node where I currently have 2 iSCSI connections and 2multipath lines (with p1p1.100 and p1p2) it is instantiated a new iSCSI SID using interface "default" and in multipath -l output I see now 3 lines
- on the newly added node I only see 1 iSCSI SID using interface default
My way to solve the situation was to go inside iscsi multipath section do nothing but save the same config
brutally on first node iscsiadm -m session -u --> all iscsi sessions are closed after a while I see again the original 2 connections recovered, with correct interface names used
- on second node iscsiadm -m session -u --> the only session is cloed nothing happens if I set to maintenance the node and then activate the node --> the 2 correct iscsi sessions are activated...
Thanks Gianluca
_______________________________________________ Users mailing list Users@ovirt.org http://lists.ovirt.org/mailman/listinfo/users

On Mon, Apr 10, 2017 at 2:12 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Gianluca,
I can see that the workaround you describe here (to complete multipath configuration in CLI) fixes an inconsistency in observed iSCSI sessions. I think it is a shortcoming in oVirt that you had to resort to manual configuration. Could you file a bug about this? Ideally, following the bug template presented to you by Bugzilla, i.e. "Expected: two iSCSI sessions", "Got: one the first node ... one the second node".
Edy, Martin, do you think you could help out here?
Thanks, Ondra
Ok, this evening I'm going to open a bugzilla for that. Please keep in mind that on the already configured node (where before node addition there were two connections in place with multipath), actually the node addition generates a third connection, added to the existing two, using "default" as iSCSI interface (clearly seen if I run "iscsiadm -m session -P1") .... Gianluca

Yes, this is what struck me about your situation. Will you be able to find relevant logs regarding multipath configuration, in which we would see when (or even why) the third connection was created on the first node, and only one connection on the second? On Mon, Apr 10, 2017 at 2:17 PM, Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
On Mon, Apr 10, 2017 at 2:12 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Gianluca,
I can see that the workaround you describe here (to complete multipath configuration in CLI) fixes an inconsistency in observed iSCSI sessions. I think it is a shortcoming in oVirt that you had to resort to manual configuration. Could you file a bug about this? Ideally, following the bug template presented to you by Bugzilla, i.e. "Expected: two iSCSI sessions", "Got: one the first node ... one the second node".
Edy, Martin, do you think you could help out here?
Thanks, Ondra
Ok, this evening I'm going to open a bugzilla for that. Please keep in mind that on the already configured node (where before node addition there were two connections in place with multipath), actually the node addition generates a third connection, added to the existing two, using "default" as iSCSI interface (clearly seen if I run "iscsiadm -m session -P1") ....
Gianluca

On Mon, Apr 10, 2017 at 2:44 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Yes, this is what struck me about your situation. Will you be able to find relevant logs regarding multipath configuration, in which we would see when (or even why) the third connection was created on the first node, and only one connection on the second?
On Mon, Apr 10, 2017 at 2:17 PM, Gianluca Cecchi < gianluca.cecchi@gmail.com> wrote:
On Mon, Apr 10, 2017 at 2:12 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Gianluca,
I can see that the workaround you describe here (to complete multipath configuration in CLI) fixes an inconsistency in observed iSCSI sessions. I think it is a shortcoming in oVirt that you had to resort to manual configuration. Could you file a bug about this? Ideally, following the bug template presented to you by Bugzilla, i.e. "Expected: two iSCSI sessions", "Got: one the first node ... one the second node".
Edy, Martin, do you think you could help out here?
Thanks, Ondra
Ok, this evening I'm going to open a bugzilla for that. Please keep in mind that on the already configured node (where before node addition there were two connections in place with multipath), actually the node addition generates a third connection, added to the existing two, using "default" as iSCSI interface (clearly seen if I run "iscsiadm -m session -P1") ....
Gianluca
vdsm log of the already configured host is here for that day: https://drive.google.com/file/d/0BwoPbcrMv8mvQzdCUmtIT1NOT2c/view?usp=sharin... Installation / configuration of the second node happened between 11:30 AM and 01:30 PM of 6th of April. Aound 12:29 you will find: 2017-04-06 12:29:05,832+0200 INFO (jsonrpc/7) [dispatcher] Run and protect: getVGInfo, Return response: {'info': {'state': 'OK', 'vgsize': '1099108974592', 'name': '5ed04196-87f1-480e-9fee-9dd450a3b53b', 'vgfree': '182536110080', 'vgUUID': 'rIENae-3NLj-o4t8-GVuJ-ZKKb-ksTk-qBkMrE', 'pvlist': [{'vendorID': 'EQLOGIC', 'capacity': '1099108974592', 'fwrev': '0000', 'pe_alloc_count': '6829', 'vgUUID': 'rIENae-3NLj-o4t8-GVuJ-ZKKb-ksTk-qBkMrE', 'pathlist': [{'connection': '10.10.100.9', 'iqn': 'iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910', 'portal': '1', 'port': '3260', 'initiatorname': 'p1p1.100'}, {'connection': '10.10.100.9', 'iqn': 'iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910', 'portal': '1', 'port': '3260', 'initiatorname': 'p1p2'}, {'connection': '10.10.100.9', 'iqn': 'iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910', 'portal': '1', 'port': '3260', 'initiatorname': 'default'}], 'pe_count': '8189', 'discard_max_bytes': 15728640, 'pathstatus': [{'type': 'iSCSI', 'physdev': 'sde', 'capacity': '1099526307840', 'state': 'active', 'lun': '0'}, {'type': 'iSCSI', 'physdev': 'sdf', 'capacity': '1099526307840', 'state': 'active', 'lun': '0'}, {'type': 'iSCSI', 'physdev': 'sdg', 'capacity': '1099526307840', 'state': 'active', 'lun': '0'}], 'devtype': 'iSCSI', 'discard_zeroes_data': 1, 'pvUUID': 'g9pjI0-oifQ-kz2O-0Afy-xdnx-THYD-eTWgqB', 'serial': 'SEQLOGIC_100E-00_64817197B5DFD0E5538D959702249B1C', 'GUID': '364817197b5dfd0e5538d959702249b1c', 'devcapacity': '1099526307840', 'productID': '100E-00'}], 'type': 3, 'attr': {'allocation': 'n', 'partial': '-', 'exported': '-', 'permission': 'w', 'clustered': '-', 'resizeable': 'z'}}} (logUtils:54) and around 12:39 you will find 2017-04-06 12:39:11,003+0200 ERROR (check/loop) [storage.Monitor] Error checking path /dev/5ed04196-87f1-480e-9fee-9dd450a3b53b/metadata (monitor:485) Traceback (most recent call last): File "/usr/share/vdsm/storage/monitor.py", line 483, in _pathChecked delay = result.delay() File "/usr/lib/python2.7/site-packages/vdsm/storage/check.py", line 368, in delay raise exception.MiscFileReadException(self.path, self.rc, self.err) MiscFileReadException: Internal file read failure: ('/dev/5ed04196-87f1-480e-9fee-9dd450a3b53b/metadata', 1, bytearray(b"/usr/bin/dd: error reading \'/dev/5ed04196-87f1-480e-9fee-9dd450a3b53b/metadata\': Input/output error\n0+0 records in\n0+0 records out\n0 bytes (0 B) copied, 0.000234164 s, 0.0 kB/s\n")) 2017-04-06 12:39:11,020+0200 INFO (check/loop) [storage.Monitor] Domain 5ed04196-87f1-480e-9fee-9dd450a3b53b became INVALID (monitor:456) that I think corresponds to the moment when I executed "iscsiadm -m session -u" and had the automaic remediation of the correctly defined paths Gianluca

On Mon, Apr 10, 2017 at 3:06 PM, Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:
On Mon, Apr 10, 2017 at 2:44 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Yes, this is what struck me about your situation. Will you be able to find relevant logs regarding multipath configuration, in which we would see when (or even why) the third connection was created on the first node, and only one connection on the second?
On Mon, Apr 10, 2017 at 2:17 PM, Gianluca Cecchi < gianluca.cecchi@gmail.com> wrote:
On Mon, Apr 10, 2017 at 2:12 PM, Ondrej Svoboda <osvoboda@redhat.com> wrote:
Gianluca,
I can see that the workaround you describe here (to complete multipath configuration in CLI) fixes an inconsistency in observed iSCSI sessions. I think it is a shortcoming in oVirt that you had to resort to manual configuration. Could you file a bug about this? Ideally, following the bug template presented to you by Bugzilla, i.e. "Expected: two iSCSI sessions", "Got: one the first node ... one the second node".
Edy, Martin, do you think you could help out here?
Thanks, Ondra
Ok, this evening I'm going to open a bugzilla for that. Please keep in mind that on the already configured node (where before node addition there were two connections in place with multipath), actually the node addition generates a third connection, added to the existing two, using "default" as iSCSI interface (clearly seen if I run "iscsiadm -m session -P1") ....
Gianluca
vdsm log of the already configured host is here for that day: https://drive.google.com/file/d/0BwoPbcrMv8mvQzdCUmtIT1NOT2c/ view?usp=sharing
Installation / configuration of the second node happened between 11:30 AM and 01:30 PM of 6th of April.
Aound 12:29 you will find:
2017-04-06 12:29:05,832+0200 INFO (jsonrpc/7) [dispatcher] Run and protect: getVGInfo, Return response: {'info': {'state': 'OK', 'vgsize': '1099108974592', 'name': '5ed04196-87f1-480e-9fee-9dd450a3b53b', 'vgfree': '182536110080', 'vgUUID': 'rIENae-3NLj-o4t8-GVuJ-ZKKb-ksTk-qBkMrE', 'pvlist': [{'vendorID': 'EQLOGIC', 'capacity': '1099108974592', 'fwrev': '0000', 'pe_alloc_count': '6829', 'vgUUID': 'rIENae-3NLj-o4t8-GVuJ-ZKKb-ksTk-qBkMrE', 'pathlist': [{'connection': '10.10.100.9', 'iqn': 'iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910', 'portal': '1', 'port': '3260', 'initiatorname': 'p1p1.100'}, {'connection': '10.10.100.9', 'iqn': 'iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910', 'portal': '1', 'port': '3260', 'initiatorname': 'p1p2'}, {'connection': '10.10.100.9', 'iqn': 'iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910', 'portal': '1', 'port': '3260', 'initiatorname': 'default'}], 'pe_count': '8189', 'discard_max_bytes': 15728640, 'pathstatus': [{'type': 'iSCSI', 'physdev': 'sde', 'capacity': '1099526307840', 'state': 'active', 'lun': '0'}, {'type': 'iSCSI', 'physdev': 'sdf', 'capacity': '1099526307840', 'state': 'active', 'lun': '0'}, {'type': 'iSCSI', 'physdev': 'sdg', 'capacity': '1099526307840', 'state': 'active', 'lun': '0'}], 'devtype': 'iSCSI', 'discard_zeroes_data': 1, 'pvUUID': 'g9pjI0-oifQ-kz2O-0Afy-xdnx-THYD-eTWgqB', 'serial': 'SEQLOGIC_100E-00_64817197B5DFD0E5538D959702249B1C', 'GUID': ' 364817197b5dfd0e5538d959702249b1c', 'devcapacity': '1099526307840', 'productID': '100E-00'}], 'type': 3, 'attr': {'allocation': 'n', 'partial': '-', 'exported': '-', 'permission': 'w', 'clustered': '-', 'resizeable': 'z'}}} (logUtils:54)
and around 12:39 you will find
2017-04-06 12:39:11,003+0200 ERROR (check/loop) [storage.Monitor] Error checking path /dev/5ed04196-87f1-480e-9fee-9dd450a3b53b/metadata (monitor:485) Traceback (most recent call last): File "/usr/share/vdsm/storage/monitor.py", line 483, in _pathChecked delay = result.delay() File "/usr/lib/python2.7/site-packages/vdsm/storage/check.py", line 368, in delay raise exception.MiscFileReadException(self.path, self.rc, self.err) MiscFileReadException: Internal file read failure: ('/dev/5ed04196-87f1-480e-9fee-9dd450a3b53b/metadata', 1, bytearray(b"/usr/bin/dd: error reading \'/dev/5ed04196-87f1-480e-9fee-9dd450a3b53b/metadata\': Input/output error\n0+0 records in\n0+0 records out\n0 bytes (0 B) copied, 0.000234164 s, 0.0 kB/s\n")) 2017-04-06 12:39:11,020+0200 INFO (check/loop) [storage.Monitor] Domain 5ed04196-87f1-480e-9fee-9dd450a3b53b became INVALID (monitor:456)
that I think corresponds to the moment when I executed "iscsiadm -m session -u" and had the automaic remediation of the correctly defined paths
Gianluca
So I come back here because I have an "orthogonal" action with the same effect. I have already in place the same 2 oVirt hosts using one 4Tb iSCSI lun. With the configuration detailed at beginning of thread, taht I resend: http://lists.ovirt.org/pipermail/users/2017-March/080992.html On them I have multipath access defined from inside oVirt like this at OS level [root@ov300 ~]# iscsiadm -m session -P 1 Target: iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910 (non-flash) Current Portal: 10.10.100.41:3260,1 Persistent Portal: 10.10.100.9:3260,1 ********** Interface: ********** Iface Name: p1p2 Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.88 Iface HWaddress: <empty> Iface Netdev: p1p2 SID: 1 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE Current Portal: 10.10.100.42:3260,1 Persistent Portal: 10.10.100.9:3260,1 ********** Interface: ********** Iface Name: p1p1.100 Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.87 Iface HWaddress: <empty> Iface Netdev: p1p1.100 SID: 2 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE So the network adapter used in the multipath config are p1p2 and p1p1.100 Now I go and add a new storage domain: a 5Tb lun on another DELL EQLOGIC storage with ip address 10.10.100.7, so on the same lan as the existing one. I go through Storage -> New Domain discover targets with the ip only one path is detected I connect and the storage domain is activated But on both oVirt hosts I have only one path for this LUN and at portal level the generic "default" Iface Name is used: iscsiadm -m session -P1 gives the 2 connections above, plus Target: iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750 (non-flash) Current Portal: 10.10.100.38:3260,1 Persistent Portal: 10.10.100.7:3260,1 ********** Interface: ********** Iface Name: default Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.88 Iface HWaddress: <empty> Iface Netdev: <empty> SID: 3 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE I can now retry to put host in maintenance and then activate and see if then it uses iscsi1 and iscsi2, but I proceed manually
From web gui Datacenter --> MyDC iSCSI Multipathing
iscsi1 -> edit I see that only the 10.10.100.9 4Tb storage target is selected --> I select the second one too (10.10.100.7) --> OK No tasks seen in web gui but at os side I see added: Current Portal: 10.10.100.37:3260,1 Persistent Portal: 10.10.100.7:3260,1 ********** Interface: ********** Iface Name: p1p1.100 Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.87 Iface HWaddress: <empty> Iface Netdev: p1p1.100 SID: 4 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE The same for iscsi2 --> edit only the 10.10.100.9 4Tb storage target is selected --> I select the second one too (10.10.100.7) --> OK Now I see for this lun a total of 3 connections (the initial "default" one and the iscsi1 and iscsi2 ones): Target: iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750 (non-flash) Current Portal: 10.10.100.38:3260,1 Persistent Portal: 10.10.100.7:3260,1 ********** Interface: ********** Iface Name: default Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.88 Iface HWaddress: <empty> Iface Netdev: <empty> SID: 3 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE ********** Interface: ********** Iface Name: p1p2 Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.88 Iface HWaddress: <empty> Iface Netdev: p1p2 SID: 5 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE Current Portal: 10.10.100.37:3260,1 Persistent Portal: 10.10.100.7:3260,1 ********** Interface: ********** Iface Name: p1p1.100 Iface Transport: tcp Iface Initiatorname: iqn.1994-05.com.redhat:f2d7fc1e2fc Iface IPaddress: 10.10.100.87 Iface HWaddress: <empty> Iface Netdev: p1p1.100 SID: 4 iSCSI Connection State: LOGGED IN iSCSI Session State: LOGGED_IN Internal iscsid Session State: NO CHANGE [root@ov300 ~]# iscsiadm -m session tcp: [1] 10.10.100.9:3260,1 iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910 (non-flash) tcp: [2] 10.10.100.9:3260,1 iqn.2001-05.com.equallogic:4-771816-e5d0dfb59-1c9b240297958d53-ovsd3910 (non-flash) tcp: [3] 10.10.100.7:3260,1 iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750 (non-flash) tcp: [4] 10.10.100.7:3260,1 iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750 (non-flash) tcp: [5] 10.10.100.7:3260,1 iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750 (non-flash) [root@ov300 ~]# At the end I manually remove the wrong "default" one created by the web gui on both nodes: [root@ov300 ~]# iscsiadm -m session -r 3 -u Logging out of session [sid: 3, target: iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750, portal: 10.10.100.7,3260] Logout of [sid: 3, target: iqn.2001-05.com.equallogic:0-8a0906-9d1c8500d-28e3c937b8d59521-ovsd3750, portal: 10.10.100.7,3260] successful. [root@ov300 ~]# Is this the expected workflow, or in case of multipathing configured at oVirt level, oVirt should have directly configured the new LUN in multipathed way? I have still to verify if I put one host in maintenance and then I activate it that all goes as expected for both LUNs. Thanks, Gianluca
participants (2)
-
Gianluca Cecchi
-
Ondrej Svoboda