Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
You could try the fall-behind feature, but it will "disconnect" (sort of) as soon as the network buffer is full, which may be very often. This may lead to your secondary being very often inconsistent, so you'd need to take some sort of snapshot before each resync. On the other hand, the slow secondary will behave much like a WAN-distant node I guess, so drbd-proxy would help - but you have to pay for it and it's not cheap. Lionel Sausin. Le 09/09/2013 18:18, Moti Levy a écrit : > I am looking for some recommendations for a specific scenario: > > Secondary DRBD node IO subsystem is busy and disk operations become > extremely slow. The secondary is still responding, but very slowly ( > for example, if you have a raid device and one drive is being rebuilt, > or if you are on amazon and their EBS volumes are degraded ). > > In this case, DRBD's queues ( application pending, pending ) start to > build up, and at some point will block ay application from writing to > the device. > > At this point if I set the primary to standalone ( or disconnect the > secondary ) the applications will regain write access. > > I am interested in having this done automatically, is there a way to > let DRBD know if application_pending becomes a write blocking issue to > disconnect and switch to standalone ? > > Moti > _______________________________________________ > drbd-user mailing list > drbd-user at lists.linbit.com > http://lists.linbit.com/mailman/listinfo/drbd-user > >