Opened 12 years ago

Closed 12 years ago

#3447 closed Bug (invalid)

Transmission Daemon crashes on low mem machine, using NFS

Reported by: spbkaizo Owned by:
Priority: Normal Milestone: None Set
Component: Daemon Version: 2.04+
Severity: Critical Keywords: nfs
Cc: simonb@…

Description

Hi All,

I've been running transmission-daemon happily on a headless machine for well over a year, tracking the svn releases. I've recently switched the filesystem to NFS, and now I've getting page allocation errors, the linux OOM killer is running rampant even though there's plenty of swap kicking around. Here's the kernel dmesg error:

[42954670.810000] transmission-da: page allocation failure. order:0, mode:0xd0 [42954670.820000] [<c0029da8>] (dump_stack+0x0/0x14) from [<c0079980>] (alloc_pages_internal+0x3c0/0x3d8) [42954670.830000] [<c00795c0>] (alloc_pages_internal+0x0/0x3d8) from [<c00799c0>] (alloc_pages+0x14/0x18) [42954670.840000] [<c00799ac>] (alloc_pages+0x0/0x18) from [<c0093d4c>] (cache_alloc_refill+0x2a0/0x57c) [42954670.850000] [<c0093aac>] (cache_alloc_refill+0x0/0x57c) from [<c0093a74>] (kmem_cache_alloc+0x4c/0x84) [42954670.860000] [<c0093a28>] (kmem_cache_alloc+0x0/0x84) from [<bf100fec>] (nfs_create_request+0x34/0x148 [nfs]) [42954670.870000] r7:00000000 r6:c031a4c0 r5:c031a4c0 r4:c1798000 [42954670.870000] [<bf100fb8>] (nfs_create_request+0x0/0x148 [nfs]) from [<bf102964>] (readpage_async_filler+0xc8/0x180 [nfs]) [42954670.890000] [<bf10289c>] (readpage_async_filler+0x0/0x180 [nfs]) from [<c007bd34>] (read_cache_pages+0x88/0xe8) [42954670.900000] [<c007bcac>] (read_cache_pages+0x0/0xe8) from [<bf102658>] (nfs_readpages+0x104/0x16c [nfs]) [42954670.910000] [<bf102554>] (nfs_readpages+0x0/0x16c [nfs]) from [<c007bb48>] (do_page_cache_readahead+0x134/0x1e8) [42954670.920000] r8:00001e00 r7:00007145 r6:c0ff96a8 r5:000015c5 r4:c1799d60 [42954670.930000] [<c007ba14>] (do_page_cache_readahead+0x0/0x1e8) from [<c007bef4>] (ondemand_readahead+0x160/0x170) [42954670.940000] [<c007bd94>] (ondemand_readahead+0x0/0x170) from [<c007bfe8>] (page_cache_sync_readahead+0x34/0x3c) [42954670.950000] r8:c0c6c8c0 r7:00005b80 r6:00000000 r5:00000000 r4:00005ba0 [42954670.960000] [<c007bfb4>] (page_cache_sync_readahead+0x0/0x3c) from [<c0075cc4>] (generic_file_aio_read+0x200/0x5a0) [42954670.970000] [<c0075ac4>] (generic_file_aio_read+0x0/0x5a0) from [<bf0f8e70>] (nfs_file_read+0xe8/0xf8 [nfs]) [42954670.980000] [<bf0f8d88>] (nfs_file_read+0x0/0xf8 [nfs]) from [<c0096f54>] (do_sync_read+0xb0/0x100) [42954670.990000] [<c0096ea4>] (do_sync_read+0x0/0x100) from [<c0097980>] (vfs_read+0xb4/0x18c) [42954671.000000] r6:c1799fa8 r5:42178000 r4:c0c6c8c0 [42954671.000000] [<c00978cc>] (vfs_read+0x0/0x18c) from [<c0097f44>] (sys_pread64+0x5c/0x80) [42954671.010000] r7:000000b4 r6:42178000 r5:c0c6c8c0 r4:00020000 [42954671.010000] [<c0097ee8>] (sys_pread64+0x0/0x80) from [<c0025aa0>] (ret_fast_syscall+0x0/0x2c) [42954671.020000] r6:00020000 r5:00000000 r4:05b80000 [42954671.030000] Mem-info: [42954671.030000] DMA per-cpu: [42954671.030000] CPU 0: hi: 0, btch: 1 usd: 0 [42954671.040000] Active:12 inactive:144 dirty:0 writeback:0 unstable:0 [42954671.040000] free:0 slab:851 mapped:1 pagetables:116 bounce:0 [42954671.050000] DMA free:0kB min:720kB low:900kB high:1080kB active:48kB inactive:576kB present:32512kB pages_scanned:136 all_unreclaimable? yes [42954671.070000] lowmem_reserve[]: 0 0 0 [42954671.070000] DMA: 0*4kB 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 0kB [42954671.080000] 167 total pagecache pages [42954671.080000] Swap cache: add 38425, delete 38421, find 12287/16509 [42954671.090000] Free swap = 97844kB [42954671.090000] Total swap = 126140kB [42954671.100000] 8192 pages of RAM [42954671.100000] 125 free pages [42954671.110000] 825 reserved pages [42954671.110000] 851 slab pages [42954671.110000] 142 pages shared [42954671.120000] 4 pages swap cached

It appears to be caused when it's re-checking data only after a crash. Clearing out all the torrents and adding them in once again from new doesn't cause a crash.

S.

Change History (4)

comment:1 Changed 12 years ago by spbkaizo

Apologies, here's a better formatted info:

[42954670.810000] transmission-da: page allocation failure. order:0, mode:0xd0
[42954670.820000] [<c0029da8>] (dump_stack+0x0/0x14) from [<c0079980>] (__alloc_pages_internal+0x3c0/0x3d8)
[42954670.830000] [<c00795c0>] (__alloc_pages_internal+0x0/0x3d8) from [<c00799c0>] (__alloc_pages+0x14/0x18)
[42954670.840000] [<c00799ac>] (__alloc_pages+0x0/0x18) from [<c0093d4c>] (cache_alloc_refill+0x2a0/0x57c)
[42954670.850000] [<c0093aac>] (cache_alloc_refill+0x0/0x57c) from [<c0093a74>] (kmem_cache_alloc+0x4c/0x84)
[42954670.860000] [<c0093a28>] (kmem_cache_alloc+0x0/0x84) from [<bf100fec>] (nfs_create_request+0x34/0x148 [nfs])
[42954670.870000]  r7:00000000 r6:c031a4c0 r5:c031a4c0 r4:c1798000
[42954670.870000] [<bf100fb8>] (nfs_create_request+0x0/0x148 [nfs]) from [<bf102964>] (readpage_async_filler+0xc8/0x180 [nfs])
[42954670.890000] [<bf10289c>] (readpage_async_filler+0x0/0x180 [nfs]) from [<c007bd34>] (read_cache_pages+0x88/0xe8)
[42954670.900000] [<c007bcac>] (read_cache_pages+0x0/0xe8) from [<bf102658>] (nfs_readpages+0x104/0x16c [nfs])
[42954670.910000] [<bf102554>] (nfs_readpages+0x0/0x16c [nfs]) from [<c007bb48>] (__do_page_cache_readahead+0x134/0x1e8)
[42954670.920000]  r8:00001e00 r7:00007145 r6:c0ff96a8 r5:000015c5 r4:c1799d60
[42954670.930000] [<c007ba14>] (__do_page_cache_readahead+0x0/0x1e8) from [<c007bef4>] (ondemand_readahead+0x160/0x170)
[42954670.940000] [<c007bd94>] (ondemand_readahead+0x0/0x170) from [<c007bfe8>] (page_cache_sync_readahead+0x34/0x3c)
[42954670.950000]  r8:c0c6c8c0 r7:00005b80 r6:00000000 r5:00000000 r4:00005ba0
[42954670.960000] [<c007bfb4>] (page_cache_sync_readahead+0x0/0x3c) from [<c0075cc4>] (generic_file_aio_read+0x200/0x5a0)
[42954670.970000] [<c0075ac4>] (generic_file_aio_read+0x0/0x5a0) from [<bf0f8e70>] (nfs_file_read+0xe8/0xf8 [nfs])
[42954670.980000] [<bf0f8d88>] (nfs_file_read+0x0/0xf8 [nfs]) from [<c0096f54>] (do_sync_read+0xb0/0x100)
[42954670.990000] [<c0096ea4>] (do_sync_read+0x0/0x100) from [<c0097980>] (vfs_read+0xb4/0x18c)
[42954671.000000]  r6:c1799fa8 r5:42178000 r4:c0c6c8c0
[42954671.000000] [<c00978cc>] (vfs_read+0x0/0x18c) from [<c0097f44>] (sys_pread64+0x5c/0x80)
[42954671.010000]  r7:000000b4 r6:42178000 r5:c0c6c8c0 r4:00020000
[42954671.010000] [<c0097ee8>] (sys_pread64+0x0/0x80) from [<c0025aa0>] (ret_fast_syscall+0x0/0x2c)
[42954671.020000]  r6:00020000 r5:00000000 r4:05b80000
[42954671.030000] Mem-info:
[42954671.030000] DMA per-cpu:
[42954671.030000] CPU    0: hi:    0, btch:   1 usd:   0
[42954671.040000] Active:12 inactive:144 dirty:0 writeback:0 unstable:0
[42954671.040000]  free:0 slab:851 mapped:1 pagetables:116 bounce:0
[42954671.050000] DMA free:0kB min:720kB low:900kB high:1080kB active:48kB inactive:576kB present:32512kB pages_scanned:136 all_unreclaimable? yes
[42954671.070000] lowmem_reserve[]: 0 0 0
[42954671.070000] DMA: 0*4kB 0*8kB 0*16kB 0*32kB 0*64kB 0*128kB 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 0kB
[42954671.080000] 167 total pagecache pages
[42954671.080000] Swap cache: add 38425, delete 38421, find 12287/16509
[42954671.090000] Free swap  = 97844kB
[42954671.090000] Total swap = 126140kB
[42954671.100000] 8192 pages of RAM
[42954671.100000] 125 free pages
[42954671.110000] 825 reserved pages
[42954671.110000] 851 slab pages
[42954671.110000] 142 pages shared
[42954671.120000] 4 pages swap cached

comment:2 Changed 12 years ago by charles

Could you please attach a gdb backtrace of Transmission's crash?

% gdb transmission-daemon
(gdb) r -f
... run until it crashes...
(gdb) thread apply all bt
... gdb prints out the backtrace, which is the "good stuff" ...
(gdb) quit
%

comment:3 Changed 12 years ago by spbkaizo

Hrm, the headline I wrote is wrong. The daemon doesn't crash, it's an issue with the kernel and nfs4 using TCP.

This issue can be closed.

comment:4 Changed 12 years ago by charles

  • Resolution set to invalid
  • Status changed from new to closed
Note: See TracTickets for help on using tickets.