| Title: | HSJ30/40 Product Conference |
| Moderator: | SSDEVO::EDMONDS |
| Created: | Mon Jul 12 1993 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 1264 |
| Total number of notes: | 4958 |
A customer running OpenVMS 5.5-2 is using host based shadowing across a
mixed architecture cluster. The disks are connected to HSJ's.
For sake of discussion the shadow set has two members each connected
to two different controller shelves. Each controller shelf is an HSJ40
pair. When a controller fails, the customer notices the load being
moved to the other HSJ in the controller shelf but the shadow set is
broken. If they do a $SHOW DEVICE DSA0: they see a shadow set with
only one member. It needs to be rejoined manually and a copy must occur.
The belief is this should be transparent to the software.
Is this normal behavior?
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 1207.1 | Timeout problem? | SSDEVO::RMCLEAN | Mon Feb 10 1997 15:22 | 1 | |
What is the timeout set to? | |||||
| 1207.2 | Specifically, what is VMS SYSGEN SHADOW_MBR_TMO? | SSDEVO::JACKSON | Jim Jackson, HSJ40 RAID team | Mon Feb 10 1997 17:45 | 0 |
| 1207.3 | 120 | PTOVAX::PEARLMAN | Tue Feb 11 1997 09:48 | 3 | |
Current
SHADOW_MBR_TMO 120
| |||||
| 1207.4 | Same problem | TOPTEN::AVERBACH | Thu Feb 13 1997 10:38 | 6 | |
I also had the same situation occur last nite while running C_SWAP
to replace a bad cache battery. Two shadow sets lost members and they
had to be manually added back in. This customer is running vms 6.1 and
SHADOW_MBR_TMO is also set to 120.
Joe
| |||||
| 1207.5 | hardware or software - any ideas? | PTOVAX::PEARLMAN | Sun Feb 16 1997 17:12 | 1 | |
| 1207.6 | EVMS::MORONEY | UHF Computers | Thu Mar 06 1997 18:05 | 6 | |
Look through the error log for the reason the shadow member being expelled. I don't know what the failover time for the HSJ40 is, but SHADOW_MBR_TIMEOUT must be longer than this time, else the member will be expelled. -Mike | |||||