Hello,

in our environment we have a "little" problem. we use an oes1-L-3node cluster. there's salvage activated. We recieve following messages in /var/log/messages:

Jan 16 10:25:48 server kernel: ndsd: page allocation failure. order:4, mode:0xd0
Jan 16 10:25:48 server kernel: Call Trace:
Jan 16 10:25:48 server kernel: [<c01572f7>] __alloc_pages+0x3a7/0x430
Jan 16 10:25:48 server kernel: [<c01573e8>] __get_free_pages+0x18/0x24
Jan 16 10:25:48 server kernel: [<c015a77a>] cache_alloc_refill+0x2da/0x5d0
Jan 16 10:25:48 server kernel: [<c015aadc>] __kmalloc+0x6c/0x80
Jan 16 10:25:48 server kernel: [<c0198592>] setxattr+0xa2/0x290
Jan 16 10:25:48 server kernel: [<c013628c>] sigprocmask+0x6c/0x100
Jan 16 10:25:48 server kernel: [<fb109744>] mpkExit+0x113/0x119 [linuxmpk]
Jan 16 10:25:48 server kernel: [<fc1c1f4e>] lsa_open+0x18f/0x1b3 [nsslsa]
Jan 16 10:25:48 server kernel: [<c01723bf>] dentry_open_it+0xef/0x220
Jan 16 10:25:48 server kernel: [<c01725e9>] filp_open+0x89/0xa0
Jan 16 10:25:48 server kernel: [<c019881a>] sys_fsetxattr+0x9a/0xf0
Jan 16 10:25:48 server kernel: [<c012eda6>] sys_time+0x16/0x50
Jan 16 10:25:48 server kernel: [<c01091e9>] sysenter_past_esp+0x52/0x79
The salvageactivated volumes show all space used (including purgeable space).
this massage appears more frequent untill finally the server hangs. Then it's only responding on pings. Neither responding on ssh nor console-inputs. The Cpu on the serveris between 98 and 80% idle, but server hangs.

For solution we deactivated salvage on those volumes an the messages disappear. And server are running fine. Any conclusions?

We won't abondon salvage.

Thanks for every advise