[vdsm] In mailbox of VDSM, the data interchange was done by 'dd' command. What would be happened once the iscsi IPs in IPSAN were unreachable?

Qixiaozhen qixiaozhen at huawei.com
Tue Aug 20 03:27:47 UTC 2013


Hi, all


1)       The type of storage domain was IPSAN.



2)       All hosts worked well in the ovirt, and the mailboxes in SPM and other hosts were OK.



3)       Something occurred in the switcher of the network, and all of the iscsi IPs in the IPSAN were unreachable.



4)       What should be happened in this case?


In my opinion, the command named 'dd' would be stuck if the data-plane in the host was off. The monitor thread of HSM and SPM would be stuck on the 'dd' command.

How to recover the VDSM in this situation? Could asynchronous IO be taken into consideration?

Operation Test:
[root at qixiaozhen tmp]#
[root at qixiaozhen tmp]# iscsiadm -m session
tcp: [1] 172.50.16.21:3260,14 iqn.2006-08.com.huawei:oceanstor:21000022a100ecba:notconfig:172.50.16.21
tcp: [2] 172.50.16.20:3260,4 iqn.2006-08.com.huawei:oceanstor:21000022a100ecba:notconfig:172.50.16.20
tcp: [3] 172.40.16.20:3260,3 iqn.2006-08.com.huawei:oceanstor:21000022a100ecba:notconfig:172.40.16.20
tcp: [4] 172.40.16.21:3260,13 iqn.2006-08.com.huawei:oceanstor:21000022a100ecba:notconfig:172.40.16.21
[root at qixiaozhen tmp]# iptables -I INPUT -s 172.50.16.21 -j DROP
[root at qixiaozhen tmp]# iptables -I INPUT -s 172.50.16.20 -j DROP
[root at qixiaozhen tmp]# iptables -I INPUT -s 172.40.16.20 -j DROP
[root at qixiaozhen tmp]# iptables -I INPUT -s 172.40.16.21 -j DROP
[root at qixiaozhen tmp]#
[root at qixiaozhen tmp]# dd if=/dev/disk/by-id/scsi-360022a110000ecba3d5728bb00000015 of=/dev/null bs=1M count=1 iflag=direct             <<<<<<<<<< iscsi ip is unreachable, operation is stuck here
^C^C^C


Thank you.

Qi
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.fedorahosted.org/pipermail/vdsm-devel/attachments/20130820/b91382dd/attachment.html>


More information about the vdsm-devel mailing list