[openstack][stein][cinder] iscsi multipath issues

Ignazio Cassano ignaziocassano at gmail.com
Wed Nov 18 16:56:08 UTC 2020


Hello, further news.The problem is not the multipath.conf.......sorry for
the mistake.The problem is the first connection from each iscsi initiator
(compute node).Seems we must start from a clean configuration where the
first 4 paths created at the first login  refer to the same disk names an
all nodes.If a node have a dirty configuration and have sdb allocated by an
iscsi path for old misconfiguration, it alloccates sdc,sdd,sde,sdf.
Another node wich start with a clean configuration, allocates
sdb,sdc,sdd,sde.So, on the first node, the first  instance allocates
sgh,sdd and so on.But If I migrate the instance to the another node, some
faulty device appears.If at the first iscsi login all node all cleaned, all
seems to work fine.

Ignazio

Il giorno mer 18 nov 2020 alle ore 15:21 Ignazio Cassano <
ignaziocassano at gmail.com> ha scritto:

> Hello I solved modifying the prio from alua to emc, and now multipath is
> cleaned when I migrate an instance.
> There is another issue:
> when I attach a new volume to the instance and then I detach the new
> volume, iscsi logout session also for the boot volume and the instance
> does non work anymore (it looses the boot volume).
> Regards
> Ignazio
>
> Il giorno mer 18 nov 2020 alle ore 12:26 Ignazio Cassano <
> ignaziocassano at gmail.com> ha scritto:
>
>> Hello Everyone,
>> I am testing stein with cinder iscsi driver for unity.
>> Since I became crazy  on queens, I decided to try with stein.
>> At this time I have only one virtual machine with only one iscsi volume.
>>
>> It was running on podiscsivc-kvm02 and I migrated it on podiscsivc-kvm01.
>>
>> Let me show what multipath -ll displays on podiscsivc-kvm02 after live
>> migration:
>>
>> 36006016006e04400dce5b45f0ac77301 dm-3 DGC     ,VRAID
>> size=40G features='1 retain_attached_hw_handler' hwhandler='1 alua' wp=rw
>> |-+- policy='round-robin 0' prio=0 status=active
>> | `- 17:0:0:60 sdm 8:192 failed faulty running
>> `-+- policy='round-robin 0' prio=0 status=enabled
>>   |- 19:0:0:60 sdk 8:160 failed faulty running
>>   `- 21:0:0:60 sdj 8:144 failed faulty running
>>
>> And now on destination node  podiscsivc-kvm01:
>> 36006016006e04400dce5b45f0ac77301 dm-3 DGC     ,VRAID
>> size=40G features='2 queue_if_no_path retain_attached_hw_handler'
>> hwhandler='1 alua' wp=rw
>> |-+- policy='round-robin 0' prio=50 status=active
>> | |- 17:0:0:14 sdm 8:192 active ready running
>> | `- 15:0:0:14 sdl 8:176 active ready running
>> `-+- policy='round-robin 0' prio=10 status=enabled
>>   |- 21:0:0:14 sdj 8:144 active ready running
>>   `- 19:0:0:14 sdk 8:160 active ready running
>>
>> On source node in /var/log/messages I get:
>> Nov 18 10:34:19 podiscsivc-kvm02 multipathd:
>> 36006016006e04400dce5b45f0ac77301: failed in domap for addition of new path
>> sdl
>> Nov 18 10:34:19 podiscsivc-kvm02 multipathd:
>> 36006016006e04400dce5b45f0ac77301: uev_add_path sleep
>> Nov 18 10:34:20 podiscsivc-kvm02 multipathd:
>> 36006016006e04400dce5b45f0ac77301: failed in domap for addition of new path
>> sdl
>> Nov 18 10:34:20 podiscsivc-kvm02 kernel: device-mapper: table: 253:3:
>> multipath: error getting device
>> Nov 18 10:34:20 podiscsivc-kvm02 kernel: device-mapper: ioctl: error
>> adding target to table
>> Nov 18 10:34:20 podiscsivc-kvm02 multipathd:
>> 36006016006e04400dce5b45f0ac77301: uev_add_path sleep
>>
>> On storage node seems it wotks fine, because the host access migrated
>> from source to destination node.
>>
>> Multipath.conf is the following:
>>
>> blacklist {
>>     # Skip the files uner /dev that are definitely not FC/iSCSI devices
>>     # Different system may need different customization
>>     devnode "^(ram|raw|loop|fd|md|dm-|sr|scd|st)[0-9]*"
>>     devnode "^hd[a-z][0-9]*"
>>     devnode "^cciss!c[0-9]d[0-9]*[p[0-9]*]"
>>
>>     # Skip LUNZ device from VNX/Unity
>>     device {
>>         vendor "DGC"
>>         product "LUNZ"
>>     }
>> }
>>
>> defaults {
>>     user_friendly_names no
>>     flush_on_last_del yes
>>     remove_retries 12
>>     skip_kpartx yes
>> }
>>
>> devices {
>>     # Device attributed for EMC CLARiiON and VNX/Unity series ALUA
>>     device {
>>         vendor "DGC"
>>         product ".*"
>>         product_blacklist "LUNZ"
>>         path_grouping_policy group_by_prio
>>         path_selector "round-robin 0"
>>         path_checker emc_clariion
>>         features "0"
>>         no_path_retry 12
>>         hardware_handler "1 alua"
>>         prio alua
>>         failback immediate
>>     }
>> }
>>
>> Why it leaves  failed faulty running devices ?
>> Is it a correct behaviour ?
>> Every time I migrate my single instance from a node to another the number
>> of files under /dev/disk/by-path increase of 4 (the number of path used for
>> unity storage) on the destination node.
>>
>>
>> I contacted Dell support and they said it is not an a problem of their
>> cinder driver but ti could be related to nova.
>> Please help me !!!
>> I must decide if acquire nfs storage or iscsi storage.
>> On the cinder compatibility drive matrix most vendors are iscsi.
>>
>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-discuss/attachments/20201118/62eebbcd/attachment.html>


More information about the openstack-discuss mailing list