cluster interconnect in Sun Fire T2000
hi,
1. Solaris OS 10 6/06 + 10_Recommended
2. two Sun Fire T2000, firmware 6.1.12, patches 119578-30, 119850-21
two crossover cables, w/o junctions
3. migration from ipge -> e1000g
4. install Sun Cluster 3.1 8/05, patches 120489-04, 120500-12
5. scinstall and scsetup OK on both nodes
6. add quorum device - OK
7. scstat -n - OK
8. scstat -q - OK
9. scstat -W - OK
10. two adapters, two cables on both nodes - OK
Problem:
If i take out crossover interconnect from e1000g2 and take it in after 10 seconds i see this in log:
Dec 12 18:37:40 kr2 e1000g: [ID 801593 kern.notice] NOTICE: pciex8086,105e - e1000g[2] : Adapter copper link is down.
Dec 12 18:37:48 kr2 cl_runtime: [ID 646950 kern.notice] NOTICE: clcomm: Path kr2:e1000g2 - kn2:e1000g2 being cleaned up
Dec 12 18:37:48 kr2 cl_runtime: [ID 489438 kern.notice] NOTICE: clcomm: Path kr2:e1000g2 - kn2:e1000g2 being drained
Dec 12 18:37:48 kr2 cl_runtime: [ID 237149 kern.notice] NOTICE: clcomm: Path kr2:e1000g2 - kn2:e1000g2 being constructed
Dec 12 18:38:49 kr2 cl_runtime: [ID 604153 kern.notice] NOTICE: clcomm: Path kr2:e1000g2 - kn2:e1000g2 errors during initiation
Dec 12 18:38:49 kr2 cl_runtime: [ID 618107 kern.warning] WARNING: Path kr2:e1000g2 - kn2:e1000g2 initiation encountered errors, errno = 62. Remote node may be down or unreachable through this path.
and scstat -W failed too now:
# scstat -W
-- Cluster Transport Paths --
EndpointEndpointStatus
----
Transport path:kn2:e1000g2 kr2:e1000g2 faulted
Transport path:kn2:e1000g1 kr2:e1000g1 Path online
this is very hard to get back the path online, reboot sometimes helps
