Closed kofemann closed 1 year ago
When NFS client issued a large number of concurrent opens to the same file, the sequence id of the open-stateid might be miscalculated. As a result, the client will get OLD_STATEID on CLOSE. The race comes from unfenced update during the open
stateid4 stateid = context .getStateHandler() .getFileTracker() .addOpen(client, owner, context.currentInode(), _args.opopen.share_access.value, _args.opopen.share_deny.value); stateid.seqid++;
When NFS client issued a large number of concurrent opens to the same file, the sequence id of the open-stateid might be miscalculated. As a result, the client will get OLD_STATEID on CLOSE. The race comes from unfenced update during the open