- Community Home
- >
- Servers and Operating Systems
- >
- Operating Systems
- >
- Operating System - HP-UX
- >
- Re: cluster lock disk issues
Categories
Company
Local Language
Forums
Discussions
Forums
- Data Protection and Retention
- Entry Storage Systems
- Legacy
- Midrange and Enterprise Storage
- Storage Networking
- HPE Nimble Storage
Discussions
Forums
Discussions
Discussions
Discussions
Forums
Discussions
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
- BladeSystem Infrastructure and Application Solutions
- Appliance Servers
- Alpha Servers
- BackOffice Products
- Internet Products
- HPE 9000 and HPE e3000 Servers
- Networking
- Netservers
- Secure OS Software for Linux
- Server Management (Insight Manager 7)
- Windows Server 2003
- Operating System - Tru64 Unix
- ProLiant Deployment and Provisioning
- Linux-Based Community / Regional
- Microsoft System Center Integration
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Discussion Boards
Community
Resources
Forums
Blogs
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-03-2002 07:18 PM
02-03-2002 07:18 PM
sorry if these have been answered before in this forum, I did some searching but couldn't really find the answers I was looking for.
am new to the MC/SG and hence these questions, please bear with me. thanx
1. it's a 2 node cluster, with one cluster lock vg/pv defined. active and passive node config.
2. if the active node has been up and running(pkgs) and cluster lock pv failed due to
harware failure(disk/path failure), would the cluster halt or panic, or we just see error mesg's in the log saying that cluster lock disk failed(I saw errors to this tune in this forum)..so am assuming as long as that active node is not shut/halted, the cluster/pkgs will keep running on active node.? lock disk is only used during cluster formation?
2. currently the single cluster lock pv is say /dev/dsk/c1t1d0 on both the active and passive nodes.
that is a cause of concern, as the access to that PV is via the same switch and FA(EMC). even though the PV has pvlink devices say for example c2t1d0 via a different switch(SAN) and FA(EMC) on both the nodes..so for best practice, in the cluster config file, for node1 use the lock disk to be /dev/dsk/c1t1d0 and for the node2 use /dev/dsk/c2t1d0. both point to the same physical lock disk via diff paths in the SAN.
3. Can we change the cluster lock PV device file on the node2 to /dev/dsk/c2t1d0..am assuming this requires shutting the cluster and then changing the config file and pusing the new changes??
please advise.
thanx in advance.
jithu
Solved! Go to Solution.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-04-2002 12:18 AM
02-04-2002 12:18 AM
SolutionIf you have a cluster that is running, and the cluster lock disc fails, the cluster will continue to run, but will log Warning messages to syslog every 60 minutes (polling interval for cluster lock disc).
If, however, one node now dies/halts/fails, the surviving node HAS to be able to get the lock, and if this is not possible then theis node will TOC, as per design.
If the cluster lock disc is unavailable at cluster formation time, the cluster will still form, but with the same issue.
The ONLY time the cluster lock disc is NEEDED is to break an exact 50% quorum issue, i.e. EXACTLY 50% of the nodes die, leaving EXACTLY 50% surviving.
If you do wish to make the change you have mentioned, ensure that the new hardware path/device you are configuring is the primary path and not the Alternate Link, as this will appear to work, but will in fact not work under a failure condition.
HTH
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-04-2002 08:10 AM
02-04-2002 08:10 AM
Re: cluster lock disk issues
"If the cluster lock disc is unavailable at cluster formation time, the cluster will still form, but with the same issue."
Does that mean, say if the cluster lock disk failed, the
node1 was shut/halted for
some reason, if I try bringing up node2(standby), it would still form cluster and bring up the packages etc even though the lock disk is un-available??
TIA
jithu
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-04-2002 09:31 AM
02-04-2002 09:31 AM
Re: cluster lock disk issues
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-04-2002 11:08 AM
02-04-2002 11:08 AM
Re: cluster lock disk issues
now am more confused:-)
1. according to the class notes/manual.
"without a cluster lock disk, the cluster would not be able to reform should one node fail, causing the entire cluster to go down"
in my case, the lock disk
was un-available, active node1
was shut/halted. tried to bring up the node2, it couldn't form the cluster and ceased cluster activities.
doesn't that mean, node2 needs
access to lock disk to bring up the packages etc..
TIA
jithu
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-04-2002 11:14 AM
02-04-2002 11:14 AM
Re: cluster lock disk issues
Hope this helps
Chris
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-04-2002 11:42 AM
02-04-2002 11:42 AM
Re: cluster lock disk issues
The case you mention is where the cluster was already runing, but when a node went down, then the other node needed to get the cluster lock disc, and failed, resulting in the remaining node going down.
Again I refer you to the fact that to create a new cluster, you must configure a cluster lock disc which must be available at the time of doing cmapplyconf. Once this is done, the cluster lock disc is ONLY used for decision making in the event of a tie, i.e. EXACTLY 50% of a cluster quorum remains.
Hence, if the cluster goes to start and the cluster lock disc is unavailable, the cluster will still form, itis only upon a failure that the cluster lock disc is actually needed.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
02-07-2002 05:34 PM
02-07-2002 05:34 PM
Re: cluster lock disk issues
I think Melvyn has explained really well on the concepts. I think you might be confused because of the terms being used.
Your post earlier reads :
1. according to the class notes/manual.
"without a cluster lock disk, the cluster would not be able to reform should one node fail, causing the entire cluster to go down"
and this is perfectly correct. The difference in terms might be causing the confusion. Here above it says the cluster will not be able to *REFORM*, and what it means is that if there is no cluster lock disk, and one of the nodes goes down ( leaving only the other running, and hence 50% of the cluster ) then the remaining node will TOC as well when it fails to get the cluster lock disk.
On the other hand, if you don't have a cluster lock disk - and you try to start a one node cluster, it will still *FORM* with or without the cluster lock disk.
Hope this helps.