Created attachment 363367 [details] /tmp/syslog Description of problem: While installing rawhide-20091001 to a IBM Power6 (eclipz) system, the installer dies at random points during stage#2. Version-Release number of selected component (if applicable): * anaconda-12.32 How reproducible: * I'm able to consistently hit this on just the power6 system, but it happens at different steps each attempt Steps to Reproduce: 1. Boot the installer using any available means on a IBM Power6 system [1] 2. Proceed with a URL install using VNC 3. Connect to VNC and continue installing, accepting all defaults [1] Include the boot argument 'nokill' in order to debug the failure after anaconda is OOM killed Actual results: VNC exits and the serial console shows ... <3>Out of memory: kill process 262 (anaconda) score 3089 or a child <3>Killed process 324 (anaconda) <4>anaconda used greatest stack depth: 4576 bytes left Expected results: No OOM Kill Additional info: * See attached syslog for more complete failure information
There are so many errors in this syslog. There are several kernel errors, plus lvm invoking OOM killer, then finally anaconda. Here's the first error, so let's start with that one and see if fixing it fixes all the others. <4>mount: page allocation failure. order:5, mode:0xc0d0 <4>Call Trace: <4>[c000000005c9b340] [c000000000014104] .show_stack+0x94/0x1c0 (unreliable) <4>[c000000005c9b3f0] [c00000000079b1e0] .dump_stack+0x28/0x3c <4>[c000000005c9b470] [c0000000001a417c] .__alloc_pages_nodemask+0x634/0x748 <4>[c000000005c9b5d0] [c0000000001df4b0] .alloc_pages_current+0xa0/0x100 <4>[c000000005c9b670] [c0000000001a1f38] .__get_free_pages+0x38/0x90 <4>[c000000005c9b700] [c0000000001ed148] .__kmalloc+0x240/0x308 <4>[c000000005c9b7d0] [c0000000002d52d0] .ext4_fill_super+0x12c8/0x2a28 <4>[c000000005c9b930] [c000000000200254] .get_sb_bdev+0x1ac/0x228 <4>[c000000005c9ba20] [c0000000002cfd48] .ext4_get_sb+0x58/0x78 <4>[c000000005c9bad0] [c0000000001ffb70] .vfs_kern_mount+0xb8/0x1d8 <4>[c000000005c9bb90] [c0000000001ffd60] .do_kern_mount+0x78/0x170 <4>[c000000005c9bc40] [c0000000002231c8] .do_mount+0x4f8/0xa78 <4>[c000000005c9bd50] [c000000000257dd4] .compat_sys_mount+0x23c/0x2f8 <4>[c000000005c9be30] [c0000000000085f0] syscall_exit+0x0/0x40
How much memory does the system have? At the top it seems to be saying it has 384MB. If that's true and you also didn't configure swap then it's hardly likely that a NUMA system with 2 nodes and 12 CPUS would install with that.
sh-4.0# cat /proc/meminfo MemTotal: 333460 kB MemFree: 10476 kB Buffers: 4632 kB Cached: 190840 kB SwapCached: 0 kB Active: 70604 kB Inactive: 100036 kB Active(anon): 65328 kB Inactive(anon): 85616 kB Active(file): 5276 kB Inactive(file): 14420 kB Unevictable: 52344 kB Mlocked: 0 kB SwapTotal: 0 kB SwapFree: 0 kB Dirty: 0 kB Writeback: 0 kB AnonPages: 27628 kB Mapped: 11112 kB Slab: 80528 kB SReclaimable: 11588 kB SUnreclaim: 68940 kB PageTables: 844 kB NFS_Unstable: 0 kB Bounce: 0 kB WritebackTmp: 0 kB CommitLimit: 166728 kB Committed_AS: 229168 kB VmallocTotal: 8589934592 kB VmallocUsed: 5568 kB VmallocChunk: 8589909420 kB HugePages_Total: 0 HugePages_Free: 0 HugePages_Rsvd: 0 HugePages_Surp: 0 Hugepagesize: 16384 kB
The specified minimum is 512MB
Yes apologies for the bz. Who stole my memory? Arghhh :(