RFC NLM lock failover
Get series with b4
b4 am [email protected]Versions newest first
- v1 RFC Latest3 patches 8 repliesIn review
Patches in v1 No applies recorded yet
| # | Subject and files | Size | Replies | Applied |
|---|---|---|---|---|
| 1/3 | NLM lock failover - lock release | +0 | 8 replies | |
| 2/3 | NLM lock failover - per ip grace period | +0 | 0 replies | |
| 3/3 | NLM lock failover = statd enablement | +0 | 0 replies |
Reviews trailers per patch
No Reviewed-by, Acked-by or Tested-by trailers for v1 yet.
Cover letter [Cluster-devel] [RFC PATCH 0/3] NLM lock failover
The uploaded patches implement NLM lock failover as discussed in:
http://www.redhat.com/archives/linux-cluster/2006-June/msg00050.html
[PATCH 1/3]
Add a new admin interface into current nfsd procfs filesystem to trigger NLM lock releasing logic. The command is invoked by echoing the server virtual IP address into /proc/fs/nfsd/nlm_unlock file as:
shell> cd /prod/fs/nfsd
shell> echo 10.10.1.1 > nlm_unlock
It is currently restricted to IPV4 addressing and the command should be invoked from the taken-over NFS server. To do list (not yet implemented) include:
1. IPv6 addessing enablement
2. Add "client:server" ip pair to allow NFS V4 lock failover as proposed by Andy Adamson (CITI).
[PATCH 2/3]
Add take-over server counter-part command into nfsd procfs interface to allow selectively setting of per (virtual) ip (lockd) grace period. It is also invoked by echoing the virtual IP into /proc/fs/nfsd/nlm_set_ip_grace file as:
shell> cd /proc/fs/nfsd
shell> echo 10.10.1.1 > nlm_set_ip_grace
It is also restricted to IPV4 addressing and the command is expected to be invoked from the take-over NFS server.
[PATCH 3/3]
This kernel patch has *not* been unit tested out yet and it needs to be paired with user mode nfs-utils changes (not ready in time for this RFC). It puts the taken-over IPv4 address in standard dot notation into the 3rd parameter of ha_callout program (see man rpc.statd for details) for "add-client" event. For "del-client", we would assume the monitored host should be removed from machine-wide lists, regardless server's interface.
Upon agree-ed on, we will integrate the changes into our cluster suite to start a full function verification test. Please comment.
-- Wendy