On Sun, Feb 15, 2015 at 5:06 PM, Tim DeNike <tim@xxxxxxxxx> wrote: > Long story short.. Ive been having problems with random files getting > uncorrectable CRC errors on a 23TB filesystem.. RAID6 on an LSI card. > Replaced all SAS cables, replaced single drives at a time, replaced > controller card. > > Finally came to the conclusion that something must be corrupt in the > CRCs as the LSI runs through its consistency checks and patrol reads > with no problems found. > > I hesitantly ran "btrfs check --init-csum-tree" with the Centos 7 > stock binaries (btrfs-progs v3.12). It ran very fast, yet > successfully... After mounting, every file had an invalid checksum and > wouldn't read. > > I already had 3.18.2 compiled in a separate path on the filesystem and > tried again with that.. I got the following output after about an hour > or so: > > [root@938el btrfs-progs]# ./btrfs check --init-csum-tree /dev/sda > Creating a new CRC tree > Checking filesystem on /dev/sda > UUID: 1de92eed-c588-4471-b853-7f6a0a22c9a6 > Reinit crc root > extent-tree.c:2657: btrfs_reserve_extent: Assertion `ret` failed. > ./btrfs[0x43d4df] > ./btrfs(btrfs_reserve_extent+0xaf8)[0x442b23] > ./btrfs(btrfs_alloc_free_block+0x57)[0x442bcb] > ./btrfs[0x43630d] > ./btrfs(btrfs_search_slot+0x130b)[0x4381ab] > ./btrfs(btrfs_csum_file_block+0x3c2)[0x446e74] > ./btrfs(cmd_check+0xf72)[0x42608e] > ./btrfs(main+0x15d)[0x4099e1] > /lib64/libc.so.6(__libc_start_main+0xf5)[0x7ffc305dfaf5] > ./btrfs[0x409599] > > > > Tried again with 3.19-rc2 and got the same thing. > > Kernel is 3.18.1 Looks similar. http://www.spinics.net/lists/linux-btrfs/msg40944.html So maybe it's worth seeing if 3.18 works and this is a regression? The .1 and .2's had pretty minor fixes though. Otherwise I'd say file a bug at bugzilla.kernel.org -- Chris Murphy -- To unsubscribe from this list: send the line "unsubscribe linux-btrfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
