liyinshubyte
liyinshubyte
> Additionally "reclaiming" locks without node failure would require some mechanism to drop the locks on ganesha1 atomically with acquiring them on ganesha2. @ffilz Hi, I'm curious about why we...
@dang Got it, thanks the reply.
@ffilz hi, currently In grace period, when client send setxattr, removexattr, setattr, rename, remove, open and lock request, ganesha will respond with NFS4ERR_GRACE. Because posix lock is advisory lock. So...
> If possible, I suggest you try and join our community calls at 7:00 AM Pacific Time on Tuesdays. @ffilz Hi, what's the link of community calls ?
@ffilz I found a case, for example, thread1:[PTHREAD_MUTEX_lock(&slot->lock)](https://github.com/nfs-ganesha/nfs-ganesha/blob/next/src/Protocols/NFS/nfs4_op_sequence.c#L118) -> [if (slot->sequence == arg_SEQUENCE4->sa_sequenceid) ](https://github.com/nfs-ganesha/nfs-ganesha/blob/next/src/Protocols/NFS/nfs4_op_sequence.c#L122)(found request replay) -> [PTHREAD_MUTEX_unlock(&slot->lock)](https://github.com/nfs-ganesha/nfs-ganesha/blob/next/src/Protocols/NFS/nfs4_op_sequence.c#L143) -> thread2: [PTHREAD_MUTEX_lock(&slot->lock)](https://github.com/nfs-ganesha/nfs-ganesha/blob/next/src/Protocols/NFS/nfs4_op_sequence.c#L118) -> thread2: [release_slot(slot)](https://github.com/nfs-ganesha/nfs-ganesha/blob/next/src/Protocols/NFS/nfs4_op_sequence.c#L188) (set cached_result to NULL) -> thread1:...
@ffilz Hi, what do you think of this patch ? https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/541868
@ffilz I submit a new patch: https://review.gerrithub.io/c/ffilz/nfs-ganesha/+/542082
@twoholes Hi, could you install tshark in client, and use `tshark -Ynfs -t ad` to see nfs-client packet, and cut and paste it here ?
@twoholes seems correct. `the size and fileId is 0.`
> I'd have to look at that much more carefully, but I think that could work. Mostly consider if there's any possible race with more than one thread racing to...