salvage - increase swap space message

[Luciano Avelino_2]

Hi managers,

I have a Alphaserver 800 5/500 running Tru64 V5.1B, the domain usr1_delphi is damaged (the disk is good, the problem is the advfs metadata), I just tried repair the domain with verify and fixfdmn but I don’t get success with it, so I am trying to recovery the domain using salvage, but it is not working.

Any body can help me ?

Tks in advance

Luciano Avelino

========================================================================

# salvage -S -D /mnt/usr1_salva_delphi -l usr1_delphi
salvage: Domain to be recovered 'usr1_delphi'
salvage: Volume(s) to be used '/dev/disk/dsk1c'
salvage: Files will be restored to '/mnt/usr1_salva_delphi'
salvage: Logfile will be placed in '/mnt/usr1_salva_delphi/salvage.log'
salvage: Starting sequential search of all volumes: 13-Dec-2005 17:31:20
salvage: Current softlimit on tag array 10000000; increase size to 83888826? y
salvage: Current softlimit on tag array 83888826; increase size to 1668115554? y
salvage: Can't allocate 13344924432 bytes.
Check for other processes (besides salvage) which are
using large amounts of memory, and kill them if
possible, or increase swap space.
If neither of those options are available, you may need
to increase max_per_proc_data_size, vm_maxvas,
and max_per_proc_address_space.
Memory fault(coredump)
# vmstat -P

Total Physical Memory = 1152.00 M
= 147456 pages

Physical Memory Clusters:

start_pfn end_pfn type size_pages / size_bytes
0 235 pal 235 / 1.84M
235 147430 os 147195 / 1149.96M
147430 147456 pal 26 / 208.00k

Physical Memory Use:

start_pfn end_pfn type size_pages / size_bytes
235 521 scavenge 286 / 2.23M
521 1532 text 1011 / 7.90M
1532 1786 data 254 / 1.98M
1786 2087 bss 301 / 2.35M
2087 2094 cfgmgmt 7 / 56.00k
2094 2095 locks 1 / 8.00k
2095 2111 pmap 16 / 128.00k
2111 2746 unixtable 635 / 4.96M
2746 2773 logs 27 / 216.00k
2773 6212 vmtables 3439 / 26.87M
6212 147430 managed 141218 / 1103.27M
============================
Total Physical Memory Use: 147195 / 1149.96M

Managed Pages Break Down:

free pages = 84619
active pages = 96
inactive pages = 5082
wired pages = 10230
ubc pages = 41477
==================
Total = 141504

WIRED Pages Break Down:

vm wired pages = 1869
ubc wired pages = 0
meta data pages = 4359
malloc pages = 3068
contig pages = 319
user ptepages = 454
kernel ptepages = 147
free ptepages = 23
==================
Total = 10239

# swapon -s
Swap partition /dev/disk/dsk7b (default swap):
Allocated space: 49152 pages (384MB)
In-use space: 1 pages ( 0%)
Free space: 49151 pages ( 99%)

Swap partition /dev/disk/dsk6c:
Allocated space: 1110845 pages (8.48GB)
In-use space: 1 pages ( 0%)
Free space: 1110844 pages ( 99%)

Swap partition /dev/disk/dsk4c:
Allocated space: 3552895 pages (27.11GB)
In-use space: 1 pages ( 0%)
Free space: 3552894 pages ( 99%)

Swap partition /dev/disk/dsk8c:
Allocated space: 36613072 pages (279.34GB)
In-use space: 1 pages ( 0%)
Free space: 36613071 pages ( 99%)


Total swap allocation:
Allocated space: 41325964 pages (315.29GB)
Reserved space: 6830 pages ( 0%)
In-use space: 4 pages ( 0%)
Available space: 41319134 pages ( 99%)
#

[Venkatesh BL · Accepted solution]

What is the size of the domain? The "13344924432" size seem to be too huge. I am wondering if some corruption in the domain's metadata is causing 'salvage' to report this erroneous message.

[Ivan Ferreira]

Try doing it in single user mode:

boot -fl s
bcheckrc
lmf reset
kloadsrv
salvage -S -D /mnt/usr1_salva_delphi -l usr1_delphi

Ensure that you have enough space to recover the files.

Check the configuration of your kernel parameters in /etc/sysconfigtab:

max_per_proc_data_size
vm_maxvas
max_per_proc_address_space

[Luciano Avelino_2]

BL,

The size of the domain is 50GB.

[Luciano Avelino_2]

Ivan,

I will try , I notice you asap.

tks

Luciano Avelino

[Juan J M Merino]

Hi Ivan Ferreira,

Luciano and I made as requested, but it continues the same problem.
You would have another suggestion?

Thanks a lot.

[Venkatesh BL]

At this stage, I suggest that you raise a HP support call and let the engineering team look at the core files.

[Luciano Avelino_2]

I increase the swap space to 500GB