Unsolved
This post is more than 5 years old
4 Posts
0
2122
June 21st, 2010 08:00
What is the rule for setting a node into "nowrite" mode
We had a disk-related issue on our centera and, during the troubleshooting of this issue, EMC support placed six of our nodes into a "nowrite" status. We asked for an explanation as to why this was done and received two answers:
- As for every node there is a threshold limit which has to be maintained so that the node should work fine. If the limit goes below 20% the node may go offline automatically
- I dialed again to the cluster and found that the used capacity on all those nodes which were set to “no write status” as true has gone above 90%(which is the permissible limit).If any of the nodes goes above 90% used capacity there is a possibility the node may go offline and it may create a Data unavailable status
I have asked EMC for further clarification on these explanations, but also figured I would ask this forum: is there a "rule of thumb" for when a node should be placed in "nowrite" status?
The explanantions provided so far concern me because we have a mixed node centera cluster and the nodes (gen3, gen4, and gen4LP) all have different storage capacities, so using a % rule could leave significantly more space unused on the larger capacity nodes.
Thanks.
cmurphynjm
4 Posts
0
June 29th, 2010 11:00
Thanks Joel. Placing these nodes in "nowrites" mode while the D-2-D copy was being performed makes sense as a precautionary step. My preference would be to take them out of this mode now that the disk copy has been completed. I'm going back and forth with EMC support on this. I appreciate your help on this issue.
srbhandary
27 Posts
1
June 29th, 2010 14:00
The node can be changes to write or no write by an EMC engineer by logging into the box without much hassle.
For more information, please refer emc id : emc111618 and : emc151211 on powerlink.
cmurphynjm
4 Posts
0
June 30th, 2010 07:00
Thanks for your response. The question I am asking is more about, "under what set of circumstances or conditions will a node will be placed into nowrites mode?" In other words, why is this decision made? We had 12 nodes placed into nowrites mode in one month while support was troubleshooting a disk issue on our cluster. Its not clear to me why these nodes were placed in this mode or if they need to remain in this mode now that the disk issue has been resolved. A centera RTS and EMC engineering are lookiing into this issue. I wanted to air this out to the ECN community to see if anyone else had a similar experience. Thanks.
srbhandary
27 Posts
1
July 6th, 2010 15:00
When a node is in nowrites (or read-only) mode it will not accept any new write, replication, restore, or regeneration requests. SDK deletes, EBR and LH updates will still be possible. A node can be put in nowrites mode by EMC Service when it
encounters problems on that node that need to be fixed. After the service intervention, EMC service will set the nowrites mode to off again.
Individual disks in the node can automatically be put in nowrites mode if the capacity on that disk has reached a predefined threshold. When sufficient capacity becomes available on that disk, the nowrites mode will automatically be set to off.
Nodes are also put in nowrite mode when set for cluster migration.
Apex_Ops
1 Rookie
•
90 Posts
0
July 29th, 2010 05:00
Hi CM,
I see you posted a question regarding no-write mode and that your question in this post appears to have been answered. Was this a correct or helpful answer? If so could you please mark the question as Answered (helpful or correct)?
Thanks for your assistance and participation in the Support Forums.
Regards,
Ronan
cmurphynjm
4 Posts
0
July 29th, 2010 06:00
Thanks everyone for your support and insights into this isssue. EMC celerra support looked into this issue and ended up placing 11 of the 14 nodes back into a read/write state. Based on discussions we have had with EMC, it sounds like the magic number for placing a node into read-only mode is 95% utilization.