Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
On Tue, May 15, 2007 at 01:46:44PM +0100, Ben Clewett wrote: > Lars, > > Lars Ellenberg wrote: > >On Tue, May 15, 2007 at 12:37:54PM +0100, Ben Clewett wrote: > >>The kernel is 2.6.16.13-4. (SUSE) > >> > >>I don't know what sort of io subsystem. I have 2x 10MBit/s NICs, one > >>dedicated for DRBD. In order to test I have throttled the bandwidth at > >>some value way larger than the speed of the card. > > > >so you have not throttled it, but you request a much higher sync rate > >than your hardware can handle. don't do that. > > There is good reason. Our production servers will be linked by fibre. > On error, like physical damage to a fibre or terminator, these have a > habit of jumping down the Ethernet protocol a couple of stops. So a > 1000baseF might become a 10baseF. The upshot is that I need to test > with this sync rate set much higher than my NIC's. > > I have now given this a value 1/10 of the ability of the NIC. But this > has not helped. Under protocol B they still jump to 'stalled' when I > write data under a resync operation. well. then use protocol C for now. if we could reproduce the behaviour you describe here, we can debug and fix it. but so far, it did not occur. -- : Lars Ellenberg Tel +43-1-8178292-0 : : LINBIT Information Technologies GmbH Fax +43-1-8178292-82 : : Vivenotgasse 48, A-1120 Vienna/Europe http://www.linbit.com : __ please use the "List-Reply" function of your email client.