dumpon on a gmirror system

[Walter Hop]

[in reply to pjd at FreeBSD.org, 20-12-2005]

Hi Pawel,

thanks for the reply!

> Are you sure you mirror partitions and not slices nor whole disks?
> Output of 'gmirror list' will be helpful.

I am mirroring the whole drive.

This is how it is setup:

virt1:~# gmirror list
Geom name: gm0
State: COMPLETE
Components: 2
Balance: round-robin
Slice: 4096
Flags: NONE
GenID: 0
SyncID: 1
ID: 2245734875
Providers:
1. Name: mirror/gm0
   Mediasize: 203928108544 (190G)
   Sectorsize: 512
   Mode: r5w5e2
Consumers:
1. Name: ad4
   Mediasize: 203928109056 (190G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Priority: 0
   Flags: DIRTY
   GenID: 0
   SyncID: 1
   ID: 3315293457
2. Name: ad6
   Mediasize: 203928109056 (190G)
   Sectorsize: 512
   Mode: r1w1e1
   State: ACTIVE
   Priority: 0
   Flags: DIRTY
   GenID: 0
   SyncID: 1
   ID: 3493568637

virt1:~# df
Filesystem         1K-blocks     Used     Avail Capacity  Mounted on
/dev/mirror/gm0s1a    253678   104876    128508    45%    /
devfs                      1        1         0   100%    /dev
/dev/mirror/gm0s1d   1012974    95224    836714    10%    /var
/dev/mirror/gm0s1e    253678       40    233344     0%    /tmp
/dev/mirror/gm0s1f 187309866 33001524 139323554    19%    /usr
procfs                     4        4         0   100%    /proc
devfs                      1        1         0   100%    /usr/opt/httpd/dev
devfs                      1        1         0   100%    /usr/opt/named/dev
   
>> How  do  people using gmirror exclusively handle the problem of saving
>> crash dumps?
>
> I, for one, prefer ddb(4), which is enough for me.

Well,  this  machine  is  colocated, so having the machine trap to the
debugger  and calling a non-trained datacenter engineer to read up the
output  isn't very tempting :) Maybe I can hook something up serially,
but it would be much better if we could just get a dump so the machine
can resume its tasks quickly.

> Other probably have stable systems and don't need dumps at all:)

That  would  be  nice :) I'm not even sure that it is a panic, but the
problem  remains  despite  hardware  replacement  so I guess something
fishy is going on...

Cheers,
Walter Hop
Transip BV

-- 
 Blaat het niet, dan schaapt het niet.