This issue is resolved upstream with https://github.com/gluster/glusterfs/issues/1925, but still not part of downstream. It was observed recently by the customer in BZ2004767. Analysis details: https://bugzilla.redhat.com/show_bug.cgi?id=2004767#c9 I would like to propose this bug to 3.5.6 considering 1. critical customer impact. 2. unusability of virtual xattrs from geo-rep.
@Prashant, dht_pt_getxattr() which was introduced by a patch (https://github.com/gluster/glusterfs/issues/405) for perf improvements in pass-through mode (i.e. where distribute-count is 1) does not handle virtual xattrs.Later on we merged the same in downstream also. For geo-rep there is a virtual getxattr done by geo-rep on the fuse mount via DHT_SUBVOL_STATUS_KEY and it must not be sent over the network. dht_getxattr() handles it but its missing in dht_pt_getxattr(), so the geo-rep session goes faulty.The issue will be observe only while dht subvol count is 1. The bug is critical for geo-rep aspect, Can you please give your vote? Thanks, Mohit Agrawal
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory (glusterfs bug fix update), and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://access.redhat.com/errata/RHBA-2022:0315