Looks good. Since you are re-using the RBD header object to send the watch notification, a running librbd client will most likely print out an error message along the lines of "failed to decode the notification" since you are sending "fsfreeze" / "fsunfreeze" as the payload, but it would be harmless. On Mon, Aug 22, 2016 at 9:13 AM, Nick Fisk <nick@xxxxxxxxxx> wrote: > Hi Jason, > > Here is my initial attempt at using the Watch/Notify support to be able to remotely fsfreeze a filesystem on a RBD. Please note this > was all very new to me and so there will probably be a lot of things that haven't been done in the best way. > > https://github.com/fiskn/rbd_freeze > > I'm not sure if calling out to bash scripts is the best way of doing the fsfreezing, but it was the easiest way I could think to > accomplish the task. And it also allowed me to fairly easily run extra checks like seeing if any files have been updated recently. > > Let me know what you think. > > Nick > >> -----Original Message----- >> From: ceph-users [mailto:ceph-users-bounces@xxxxxxxxxxxxxx] On Behalf Of Nick Fisk >> Sent: 08 July 2016 09:58 >> To: dillaman@xxxxxxxxxx >> Cc: 'ceph-users' <ceph-users@xxxxxxxxxxxxxx> >> Subject: Re: RBD Watch Notify for snapshots >> >> Thanks Jason, >> >> I think I'm going to start with a bash script which SSH's into the machine to check if the process has finished writing and then > calls the >> fsfreeze as I've got time constraints to getting this working. But I will definitely revisit this and see if there is something I > can create >> which will do as you have described, as it would be a much neater solution. >> >> Nick >> >> > -----Original Message----- >> > From: ceph-users [mailto:ceph-users-bounces@xxxxxxxxxxxxxx] On Behalf >> > Of Jason Dillaman >> > Sent: 08 July 2016 04:02 >> > To: nick@xxxxxxxxxx >> > Cc: ceph-users <ceph-users@xxxxxxxxxxxxxx> >> > Subject: Re: RBD Watch Notify for snapshots >> > >> > librbd pseudo-automatically handles this by flushing the cache to the >> > snapshot when a new snapshot is created, but I don't think krbd does the same. If it doesn't, it would probably be a nice > addition to >> the block driver to support the general case. >> > >> > Baring that (or if you want to involve something like fsfreeze), I >> > think the answer depends on how much you are willing to write some >> > custom C/C++ code (I don't think the rados python library exposes >> > watch/notify APIs). A daemon could register a watch on a custom per-host/image/etc object which would sync the disk when a >> notification is received. Prior to creating a snapshot, you would need to send a notification to this object to alert the daemon > to >> sync/fsfreeze/etc. >> > >> > On Thu, Jul 7, 2016 at 12:33 PM, Nick Fisk <mailto:nick@xxxxxxxxxx> wrote: >> > Hi All, >> > >> > I have a RBD mounted to a machine via the kernel client and I wish to >> > be able to take a snapshot and mount it to another machine where it can be backed up. >> > >> > The big issue is that I need to make sure that the process writing on >> > the source machine is finished and the FS is sync'd before taking the snapshot. >> > >> > My question. Is there something I can do with Watch/Notify to trigger >> > this checking/sync process on the source machine before the snapshot is actually taken? >> > >> > Thanks, >> > Nick >> > >> > _______________________________________________ >> > ceph-users mailing list >> > mailto:ceph-users@xxxxxxxxxxxxxx >> > http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com >> > >> > >> > >> > >> > -- >> > Jason >> >> _______________________________________________ >> ceph-users mailing list >> ceph-users@xxxxxxxxxxxxxx >> http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com > -- Jason _______________________________________________ ceph-users mailing list ceph-users@xxxxxxxxxxxxxx http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com
 |