iscsiadm: iface iter could not read dir /var/lib/iscsi/nodes/

[bharatv...@gmail.com]

Dear All,

I was trying to login to a SAN Device logical volume from my linux server, which is connected to SAN Device with 10Gb NIC card.

IQN of volume was : iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0018.target0016 and I executed command as follows:

"/sbin/iscsiadm -m node -T iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0018.target0016 -p 172.168.2.165 --login"

But, it failed with:

Error Code : 6

Error message : iscsiadm: iface iter could not read dir /var/lib/iscsi/nodes/iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0048.target0012/172.168.2.165,3260,3.

iscsiadm: Could not execute operation on all records: encountered iSCSI database failure

This error message was quite confusing as it shows that reading iface for IQN : "iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0048.target0012" failed but I tried login to "iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0018.target0016".

Moreover, iSCSI session for "iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0048.target0012" was logged out around 6 minutes ago.

I am unaware about the iscsi internals, and could not understand the reason for it.

I suspect that at the time of iSCSI login all iface are read.

Please help me in this.

Thanks in anticipation.

[Mike Christie]

On 12/26/2014 07:20 AM, bharatv...@gmail.com wrote:
> Dear All,
>
> I was trying to login to a SAN Device logical volume from my linux
> server, which is connected to SAN Device with 10Gb NIC card.
> IQN of volume was
> : iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0018.target0016
> and I executed command as follows:
>
> "/sbin/iscsiadm -m node -T
> iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0018.target0016
> -p 172.168.2.165 --login"
>
> But, it failed with:
> Error Code : 6
> Error message : iscsiadm: iface iter could not read dir
> /var/lib/iscsi/nodes/iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0048.target0012/172.168.2.165,3260,3.
> iscsiadm: Could not execute operation on all records: encountered iSCSI
> database failure
>
> This error message was quite confusing as it shows that reading iface
> for IQN :
> "iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0048.target0012"
> failed but I tried login to
> "iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0018.target0016".
> Moreover, iSCSI session for
> "iqn.2001-03.jp.nec:storage01:ist-m000-sn-0000000942014090.lx-ddsldset-0048.target0012"
> was logged out around 6 minutes ago.
>

Can you easily replicate this? If so, could you run iscsiadm with
debugging on? Just add "-d 8":


/sbin/iscsiadm -m node -T
iqn.2001-03.jp.nec:storage01:ist-m000-sn0000000942014090.lx-ddsldset-0018.target0016
-p 172.168.2.165 --login -d 8

[bharatv...@gmail.com]

Can you easily replicate this? If so, could you run iscsiadm with
debugging on? Just add "-d 8":


/sbin/iscsiadm -m node -T
iqn.2001-03.jp.nec:storage01:ist-m000-sn0000000942014090.lx-ddsldset-0018.target0016
-p 172.168.2.165 --login -d 8

Hi Mike,

This problem is intermittent. I can try reproducing this, but i am not sure whether this problem would hit or not.

Is there some particular condition in which "iscsiadm: Could not execute operation on all records: encountered iSCSI database failure" is retrieved ? 

Is it safe to create multiple parallel login session's by iscsiadm command ?

Eg:

 /sbin/iscsiadm -m node -T <IQN-1> -p <PORT> --login &

 /sbin/iscsiadm -m node -T <IQN-2> -p <PORT> --login &

 /sbin/iscsiadm -m node -T <IQN-3> -p <PORT> --login &

 /sbin/iscsiadm -m node -T <IQN-4> -p <PORT> --login &

..

..

 /sbin/iscsiadm -m node -T <IQN-20> -p <PORT> --login &

 and what can be consequences of logging into the same IQN again and again without logout from it?

[Dhiraj Surana]

hi , 

did you get the resolution for this issue , even i am seeing hte similar kind of issue while deleting and creating the iscsi and iser session.

20210903 10:44:36 [ 11175] run_system_cmd: Running: iscsiadm -m node --targetname iqn.1986-03.com.ibm:2145.stand2fab3plus9.118.54.153.node1 -I iface.stand2host1:1 -p 192.170.15.10:3260 --login iscsiadm: Could not execute operation on all records: encountered iSCSI database failure 20210903 10:44:36 [ 11175] run_system_cmd: rc=6, signal=0 core_available=0 when running iscsiadm -m node --targetname iqn.1986-03.com.ibm:2145.stand2fab3plus9.118.54.153.node1 -I iface.stand2host1:1 -p 192.170.15.10:3260 --login, 20210903 10:44:36 [ 11175] ------------------------------------------------------------------ run_system_cmd: iscsiadm -m node --targetname iqn.1986-03.com.ibm:2145.stand2fab3plus9.118.54.153.node1 -I iface.stand2host1:1 -p 192.170.15.10:3260 --login failed, rc=6, signal=0, core_available=0 - exiting.... 20210903 10:44:36 [ 11175] ------------------------------------------------------------------

[root@stand2host1 nodes]# iscsiadm -m node -o delete

iscsiadm: Could not execute operation on all records: encountered iSCSI database failure

[The Lee-Man]

Hi!

My apologies to bharatvivek2972. I believe Mike, who was answering questions here, moved on from open-iscsi, and I didn't notice this thread needed more attention.

I believe the problem here is that iscsiadm is not set up for parallel operation. So no, you can not run logins in parallel.

But, there is a new "no wait" option that one could pass to iscsiadm that would speed up such serialized requests. It's the "-W"/"--no_wait". Your distro may not have this feature yet, but it's upstream, and it tells iscsiadm not to wait for the login to complete. When the caller uses this option, iscsiadm returns as soon as it has sent the login request to the target. The login either fails or succeeds, in the background, as managed by iscsid. It's up to the caller to poll or check for success. Using this option, one could do:

/sbin/iscsiadm -m node -T <IQN-1> -p <PORT> --login -W

 /sbin/iscsiadm -m node -T <IQN-2> -p <PORT> --login -W

 /sbin/iscsiadm -m node -T <IQN-3> -p <PORT> --logiDhirajn -W

 /sbin/iscsiadm -m node -T <IQN-4> -p <PORT> --login -W

Of course, it might be faster to run:

/sbin/iscsiadm -m node -l -W

i.e. let iscsiadm log into all the targets in the node database, but in the case where you only want some of the targets to be logged into, the above should work.

In your case Dhiraj I'm not sure you are doing things in parallel. Are you? It might be that the node database was screwed up by some earlier parallel operation(s)?