2017-04-02 16:34 GMT-04:00 Florian Gleixner
On 02.04.2017 21:56, Ciro Iriarte wrote:
Hi!, can anyone comment on XFS stability on Leap 42.2 using crc=1?. I've read that crc=1 should be the new default. I created a XFS filesystem with explicit crc=1.
Everything was working fine until I used that filesystem as a NFS repository for ESXi, tried to move some VMs to it and had a complete hang, hours later the filesystem went to read only mode.
Trying to repair the FS I get CRC errors on the pending changelog. I can purge the log, as the entries reference the directories of the VMs I tried to move but I'm a little worried about keeping it like that. I've just moved from BTRFS trying to avoid this issues :/
Regards,
Just checked: crc=1 is default on Leap 42.2. I have some machines using XFS on Leap 42.2 without problems. Have you checked the logs and dmesg? You probably have a broken disk.
The weird thing is that according the man page, the default should be crc=0. Currently the FS sits on top of a dmraid array reshaped from R5 to R6 (still in progress). Current messages complain with: 13780.768541] XFS (md4): log record CRC mismatch: found 0x759a7a5e, expected 0x6131115a. [13780.768546] ffffc90002040200: 00 00 00 01 00 00 00 00 69 01 00 00 40 aa fd a7 ........i...@... [13780.768547] ffffc90002040210: 00 00 00 10 69 00 00 00 4e 41 52 54 28 00 00 00 ....i...NART(... The dmesg errors from the hang event where already purged (had a reboot since then). I'm waiting for the reshape to finish to do any further maneuver, but I would like to know if disabling CRC is good idea (or even an option). Regards, -- Ciro Iriarte http://iriarte.it -- -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org