Date: Thu, 13 Jun 2013 11:56:27 -0400 From: Jona Schuman <jonaschuman@gmail.com> To: Ivailo Tanusheff <Ivailo.Tanusheff@skrill.com> Cc: "freebsd-fs@freebsd.org" <freebsd-fs@freebsd.org> Subject: Re: zfs send/recv dies when transferring large-ish dataset Message-ID: <CAC-LZTajW0SO_dH9ZtUH80zX628vcosL_vOzwkwB1JF1PZy0qA@mail.gmail.com> In-Reply-To: <57e0551229684b69bc27476b8a08fb91@DB3PR07MB059.eurprd07.prod.outlook.com> References: <CAC-LZTYLzFPTvA6S4CN0xTd-E_x9c3kxYwQoFed5LkVBrwVk0Q@mail.gmail.com> <57e0551229684b69bc27476b8a08fb91@DB3PR07MB059.eurprd07.prod.outlook.com>
next in thread | previous in thread | raw e-mail | index | archive | help
machine2# nc -d -l 9999 | zfs receive -v -F -d storagepool machine1# zfs send -v -R dataset@snap | nc machine2 9999 machine1-output: sending from @ to dataset@snap machine2-output: receiving full stream of dataset@snap into storagepool/dataset@snap machine1-output: warning: cannot send 'dataset@snap': Broken pipe machine1-output: Broken pipe On Thu, Jun 13, 2013 at 3:42 AM, Ivailo Tanusheff <Ivailo.Tanusheff@skrill.com> wrote: > Hi, > > Can you try send/recv with the -v or with -vP swiches, so you can see mor= e verbose information? > > Regards, > Ivailo Tanusheff > > -----Original Message----- > From: owner-freebsd-fs@freebsd.org [mailto:owner-freebsd-fs@freebsd.org] = On Behalf Of Jona Schuman > Sent: Thursday, June 13, 2013 2:41 AM > To: freebsd-fs@freebsd.org > Subject: zfs send/recv dies when transferring large-ish dataset > > Hi, > > I'm getting some strange behavior from zfs send/recv and I'm hoping someo= ne may be able to provide some insight. I have two identical machines runni= ng 9.0-RELEASE-p3, each having a ZFS pool (zfs 5, zpool > 28) for storage. I want to use zfs send/recv for replication between the = two machines. For the most part, this has worked as expected. > However, send/recv fails when transferring the largest dataset (both in a= ctual size and in terms of number of files) on either machine. > With these datasets, issuing: > > machine2# nc -d -l 9999 | zfs recv -d storagepool machine1# zfs send data= set@snap | nc machine2 9999 > > terminates early on the sending side without any error messages. The rece= iving end continues on as expected, cleaning up the partial data received s= o far and reverting to its initial state. (I've tried using mbuffer instead= of nc, or just using ssh, both with similar results.) Oddly, zfs send dies= slightly differently depending on how the two machines are connected. When= connected through the racktop switch, zfs send dies quietly without any in= dication that the transfer has failed. > When connected directly using a crossover cable, zfs send dies quietly an= d machine1 becomes unresponsive (no network, no keyboard, hard reset requir= ed). In both cases, no messages are printed to screen or to anything in /va= r/log/. > > > I can transfer the same datasets successfully if I send/recv to/from file= : > > machine1# zfs send dataset@snap > /tmp/dump machine1# scp /tmp/dump machi= ne2:/tmp/dump machine2# zfs recv -d storagepool < /tmp/dump > > so I don't think the datasets themselves are the issue. I've also success= fully tried send/recv over the network using different network interfaces (= 10GbE ixgbe cards instead of the 1GbE igb links), which would suggest the i= ssue is with the 1GbE links. > > Might there be some buffering parameter that I'm neglecting to tune, whic= h is essential on the 1GbE links but may be less important on the faster li= nks? Are there any known issues with the igb driver that might be the culpr= it here? Any other suggestions? > > Thanks, > Jona > _______________________________________________ > freebsd-fs@freebsd.org mailing list > http://lists.freebsd.org/mailman/listinfo/freebsd-fs > To unsubscribe, send any mail to "freebsd-fs-unsubscribe@freebsd.org" > >
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CAC-LZTajW0SO_dH9ZtUH80zX628vcosL_vOzwkwB1JF1PZy0qA>