disk commit failed with error : cannot save CouchDB docs for vbucket = 77 when loading 100M items blobs

Perhaps some race because it was working in my previous run with exactly same code.

I was running some single node write throughput experiment. With goal of loading 100M short (~256 bytes) blobs into single node 128 vbuckets cluster.

Everything starts like this:

[ns_server:debug,2012-08-15T20:00:21.224,n_10@127.0.0.1:<0.6728.0>:ns_pubsub:do_subscribe_link:149]Deleting

{ns_stats_event,<0.742.0>} event handler: ok
[ns_server:debug,2012-08-15T20:00:21.291,n_10@127.0.0.1:<0.6734.0>:ns_pubsub:do_subscribe_link:120]Started subscription {ns_stats_event,<0.742.0>}

[ns_server:debug,2012-08-15T20:00:22.224,n_10@127.0.0.1:<0.6734.0>:ns_pubsub:do_subscribe_link:149]Deleting

{ns_stats_event,<0.742.0>} event handler: ok
[ns_server:debug,2012-08-15T20:00:22.266,n_10@127.0.0.1:<0.6743.0>:ns_pubsub:do_subscribe_link:120]Started subscription {ns_stats_event,<0.742.0>}

[ns_server:info,2012-08-15T20:00:22.972,n_10@127.0.0.1:ns_port_memcached:ns_port_server:log:169]memcached<0.769.0>: 537 3: Warning: failed to save docs to database, error=error reading file errno=Resource temporarily unavailable
memcached<0.769.0>: 537 3: Warning: commit failed, cannot save CouchDB docs for vbucket = 77 rev = 2
memcached<0.769.0>: 537 3: Fatal error in persisting SET ``10001711'' on vb 77!!! Requeue it...
memcached<0.769.0>: 537 3: Fatal error in persisting SET ``10002964'' on vb 77!!! Requeue it...
memcached<0.769.0>: 537 3: Fatal error in persisting SET ``10015799'' on vb 77!!! Requeue it...

And then it doesn't seem to recover.