Please don't reply to lustre-devel. Instead, comment in Bugzilla by using the following link: https://bugzilla.lustre.org/show_bug.cgi?id=12769
(In reply to comment #17) > > There are no such guarantee on any fs ever. Consider a case that you wrote > this > data on local fs, data went into cache, no error returned. Now you poweroff > your > machine - no data written. What you describe is quite similar. > If you need to make sure your data made it to the disk - you need to do some > sort of sync/fsync after you wrote your data. > Hope this makes it a little bit more clear. Is it possible that we need to challenge some assumptions when it comes to Lustre on WAN, especially configurations with higher latency networks? Perhaps we're experiencing evidence of a higher frequency use case that we're not presently handling very well? _______________________________________________ Lustre-devel mailing list [email protected] https://mail.clusterfs.com/mailman/listinfo/lustre-devel
