This post is more than 5 years old
12 Posts
0
4207
June 14th, 2018 23:00
New VMware host says network problems, VNX says it's logged in. Still no datastores
Good morning, all!
I'm running in to a problem I haven't seen before. I want to add a new VMware host to my cluster and attach it to my VNX 5300 array. I set up the software initiator, pointed it to the IP addresses on the array, and rescanned all HBAs and datastores. The scan returned one datastore of 0 bytes; not the five datastores of about 6.5 TB I was expecting.
Worked with VMware support to verify the setups; the networking and storage is set up correctly. The error I get in the host logs is something like this:
2018-06-14T12:24:21.518Z: [iscsiCorrelator] 10190302461us: [vob.iscsi.target.connect.error] vmhba41 @ vmk2 failed to login to iqn.1992-04.com.emc:cx.apm00111800838.b6 because of a network connection failure.
2018-06-14T12:24:21.518Z: [iscsiCorrelator] 10190302640us: [esx.problem.storage.iscsi.target.connect.error] Login to iSCSI target iqn.1992-04.com.emc:cx.apm00111800838.b6 on vmhba41 @ vmk2 failed. The iSCSI initiator could not establish a network connection to the target.
2018-06-14T12:24:21.518Z: [iscsiCorrelator] 10190302829us: [vob.iscsi.target.connect.error] vmhba41 @ vmk2 failed to login to iqn.1992-04.com.emc:cx.apm00111800838.b7 because of a network connection failure.
2018-06-14T12:24:21.518Z: [iscsiCorrelator] 10190303096us: [esx.problem.storage.iscsi.target.connect.error] Login to iSCSI target iqn.1992-04.com.emc:cx.apm00111800838.b7 on vmhba41 @ vmk2 failed. The iSCSI initiator could not establish a network connection to the target.
From what I see, the initiator couldn't make a connection to the target due to a network problem. However, I could vmkping through the vnics on the storage vSwitch to each of the targets. Additionally, Unisphere showed the host as connected and logged in. The difference between the new host and the old, connected hosts is that the new host shows a status "Not connected to host agent" and the Storage Group shown for the old hosts is "bluemound ESX" and the new host just shows "~management." Oh, and we're not using CHAP.
I haven't seen this behavior before - if the networking is good and there's no CHAP, the array presents its resources to anyone who asks.
Any thoughts on what might be happening?
Thanks to all for looking!
Gregg
kelleg
4.5K Posts
0
June 15th, 2018 09:00
When you add a new host to an existing array and you want to add that host to an existing Storage Group, you need to add the host to the Storage Group. The ~management means the host is connected to the array but by default it gets added to a non-storage group entity so you need to add it to the Storage Group in order to see the LUNs. Or you can create a new Storage Group and add the new host and the existing LUNs to that new Storage Group.
Got to (IE seems to work better then Chrome):
https://mydocuments.emc.com/
Then select VNX, then "VNX Server Tasks", then "Attach a Server"
You need to also check that you've properly disabled the TCP Delayed ACK on the host (this should be done for all hosts connecting to an EMC array when using iSCSI) - see VMware KB
http://kb.vmware.com/selfservice/documentLink.do?externalID=1002598
gregg2238
12 Posts
0
June 15th, 2018 14:00
Hello, Glen!
Thanks for this - it worked fine! I'm migrating away from the VNX arrays; they're being repurposed to another stack. I'll watch for the TCP Delayed Ack issue on that stack.
Thanks!
Gregg
ronw1309
2 Posts
0
August 24th, 2018 03:00
From what I see, the initiator couldn't make a connection to the target due to a network problem. However, I could vmkping through the vnics on the storage vSwitch to each of the targets. Additionally, Unisphere showed the host as connected and logged in. The difference between the new host and the old, connected hosts is that the new host shows a status "Not connected to host agent" and the Storage Group shown for the old hosts is "bluemound ESX" and the new host just shows "~management." Oh, and we're not using CHAP.