hi Kevin,
Welcome! I just checked client3_3_lookup code. I see that it
copies gfid from loc->parent if it exists and non-null. Otherwise picks
it from loc->pargfid. So I am not understanding why this patch fixed the
problem :-(. And the log you pasted in commit message fails with EINVAL.
Does NFS convert EINVAL to EROFS? Could you give more information about
how and when this happens? If you have any logs that can throw more
light on the problem feel free to share it with us.
Pranith
On 03/21/2016 11:42 PM, Kevin Vigor wrote:
Hi,
I've recently started at Facebook and will be working on glusterfs
with Richard, Shreyas and others you already know. I hope to get to
know you all soon, and not to pester you too much with dumb questions.
With that in mind, I'd appreciate eyes on
http://review.gluster.org/#/c/13800/
<http://review.gluster.org/#/c/13800/> . This small change seems to
fix (very) intermittent EROFS errors observed when files are written
during a heal of a replicated volume. However, as a newbie, I'm not
super-confident that this is the proper fix. Any help appreciated.
Thanks,
Kevin
_______________________________________________
Gluster-devel mailing list
[email protected]
http://www.gluster.org/mailman/listinfo/gluster-devel
_______________________________________________
Gluster-devel mailing list
[email protected]
http://www.gluster.org/mailman/listinfo/gluster-devel