After playing with failover, I ended with a file that cannot be read anymore. I always get EIO. glusterfs client log is as below. Is there a trick to trigger self-heal again? [2011-06-02 09:57:36.471934] I [afr-common.c:735:afr_lookup_done] 0-gfs1: background meta-data self-heal triggered. path: /netbsd/usr/src/share/mk/sys.mk [2011-06-02 09:57:36.475313] I [client3_1-fops.c:890:client3_1_getxattr_cbk] 0-hangar1: remote operation failed: Operation not supported [2011-06-02 16:07:57.33910] E [afr-self-heal-metadata.c:521:afr_sh_metadata_fix] 2-gfs1: Unable to self-heal permissions/ownership of '/netbsd/usr/src/share/mk/sys.mk' (possible split-brain). Please fix the file on all backend volumes [2011-06-02 16:07:57.34515] I [afr-self-heal-metadata.c:81:afr_sh_metadata_done] 2-gfs1: aborting selfheal of /netbsd/usr/src/share/mk/sys.mk OPEN() /netbsd/usr/src/share/mk/sys.mk => -1 (Input/output error) [2011-06-02 16:07:57.35804] W [fuse-bridge.c:582:fuse_fd_cbk] 0-glusterfs-fuse: 13867671: OPEN() /netbsd/usr/src/share/mk/sys.mk => -1 (Input/output error) -- Emmanuel Dreyfus http://hcpnet.free.fr/pubz manu@xxxxxxxxxx
