| Title: | LSM |
| Moderator: | SMURF::SHIDERLY |
| Created: | Mon Jan 17 1994 |
| Last Modified: | Fri Jun 06 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 803 |
| Total number of notes: | 2852 |
I have a customer who's system (one half of a DECsafe ASE cluster) failed
to save a crash dump after a forced crash. The error message produced at
boot time was:
savecore: Can't set LSM swap volume preference in
/dev/vol/rootdg/swapvol
From info in other notes in this conference I believe this can be caused
by three situations:
1. No rootvol volume
2. With multiple swap devices (and LSM) /dev/vol/rootdg/swapvol isn't
the primary swap device.
3. The minor number for the device file /dev/vol/rootdg/swapvol isn't
a value of 1.
But none of these situations fit the customer system. I got them to send
in the output from the commands:
# file /dev/vol/rootdg/swapvol
..result.. /dev/vol/rootdg/swapvol: block special (40/1)
# volprint -vht
# voldisk list
..results..
# volprint -vht
V NAME USETYPE KSTATE STATE LENGTH READPOL PREFPLEX
PL NAME VOLUME KSTATE STATE LENGTH LAYOUT ST-WIDTH
MODE
SD NAME PLEX PLOFFS DISKOFFS LENGTH DISK-NAME
DEVICE
v rootvol root ENABLED ACTIVE 300000 ROUND -
pl rootvol-01 rootvol ENABLED ACTIVE 300000 CONCAT -
RW
sd rz0-01 rootvol-01 0 299984 16 rz0 rz0
sd rz0-02 rootvol-01 16 0 299984 rz0 rz0
pl rootvol-02 rootvol ENABLED ACTIVE 300000 CONCAT -
RW
sd rz1-01 rootvol-02 0 299984 16 rz1 rz1
sd rz1-02 rootvol-02 16 0 299984 rz1 rz1
v swapvol swap ENABLED ACTIVE 2047472 ROUND -
pl swapvol-01 swapvol ENABLED ACTIVE 2047472 CONCAT -
RW
sd rz0b-01 swapvol-01 0 0 2047472 rz0b rz0b
pl swapvol-02 swapvol ENABLED ACTIVE 2047472 CONCAT -
RW
sd rz1b-01 swapvol-02 0 0 2047472 rz1b rz1b
v volrz0d gen ENABLED ACTIVE 1655984 SELECT -
pl volrz0d-01 volrz0d ENABLED ACTIVE 1655984 CONCAT -
RW
sd advfs_rz0d-01 volrz0d-01 0 0 1655984 advfs_rz0d
rz0d
pl volrz0d-02 volrz0d ENABLED ACTIVE 1655984 CONCAT -
RW
sd disk1d-01 volrz0d-02 0 0 1655984 disk1d rz1d
..and..
# voldisk list
DEVICE TYPE DISK GROUP STATUS
rz0 sliced rz0 rootdg online
rz0b nopriv rz0b rootdg online
rz0d nopriv advfs_rz0d rootdg online
rz0h simple rz0h rootdg online
rz1 sliced rz1 rootdg online
rz17 sliced - - offline
rz18 sliced - - offline
rz19 sliced - - offline
rz1b nopriv rz1b rootdg online
rz1d nopriv disk1d rootdg online
rz1h simple disk01 rootdg online
rz20 sliced - - offline
rz25 sliced - - offline
rz26 sliced - - offline
rz27 sliced - - offline
rz28 sliced - - offline
rz29 sliced - - offline
rz37 sliced - - offline
Anyone got any suggestions? I'm not sure where to go from here.
Regards,
John
--------------------------------------------------------
John McNulty | Email: [email protected]
UK CSC, Unix Support Group | Tel: (44) 1256 373862
Digital Equipment Corporation | DTN: 833-3862
[Posted by WWW Notes gateway]
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 794.1 | Volume Read Policy? | NNTPD::"[email protected]" | John McNulty | Mon May 12 1997 09:26 | 8 |
Could the volume read policy have anything to do with this? I've asked the customer to send me the output from # volprint -lg rootdg. And if the read policy is ROUND rather than PREFER, is it possible that the dump was written to one plex, but savecore (when setting VOL_SET_SWAP_PREF) picked the wrong one to read from? John [Posted by WWW Notes gateway] | |||||
| 794.2 | Can No One Help Me? | NNTPD::"[email protected]" | John McNulty | Fri May 16 1997 11:03 | 5 |
Come on folks, someone must be able to shed some light on this. John [Posted by WWW Notes gateway] | |||||
| 794.3 | KITCHE::schott | Eric R. Schott USG Product Management | Sun May 18 1997 21:11 | 6 | |
Hi If you require this for a customer, you should file a CLD/IPMT. This conference is not a support mechanism... | |||||
| 794.4 | NNTPD::"[email protected]" | John McNulty | Mon May 19 1997 07:46 | 21 | |
> If you require this for a customer, you should file a CLD/IPMT.
>
> This conference is not a support mechanism...
I'm aware of that. But getting an answer in NOTES (if there is one to
be had) is an order of magnitude more cost effective than raising an
IPMT. To raise an IPMT for this, I will have to send either Checkit
or UniCensus to the customer, get him to run and post back the results.
Hours of work, postage costs and days of delay. Then it takes an hour
or so to properly write up the IPMT, and then requires the involvement
of at least two administrators to log, transfer, and then route the
IPMT (one at my end, one at yours) and manage the dialogue between the
engineer and myself. I've no idea what kind of paper work or hoops your
engineers then have to jump through to work the call. Plus I believe
that Engineering are already swamped with more IPMT's than they would
wish to have.
So please excuse me if I try to exhaust all possibilites before finally
resorting to an IPMT.
[Posted by WWW Notes gateway]
| |||||
| 794.5 | rc.config? | PARVAX::JANOCK | Wed Jun 04 1997 16:38 | 3 | |
did you check to ensure the rc.config value for swap
was modified after setting up swapvol? -jeff
| |||||