Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi, > Either return as quickly as possible, > or do not return at all. > > Either way, don't call back into DRBD kernel code > from inside a synchronously running handler. Nowhere does it state that DRBD will detach automatically in all cases, but only if 'detach' strategy is set. Furthermore, the manual states: "It is entirely left to the administrator's discretion to implement I/O error handling using the command (or script) invoked by local-io-error." Nowhere does it say, don't detach manually or call DRBD kernel code. If I want to implement I/O error handling in the way: notify me and detach (not necessarily in this order), how would I proceed? > It won't be "a bit out of sync". > > It will be *inconsistent*. > That's a polite word for "corrupt by design". Doesn't DRBD preserve write order? So if one node is Ahead and the other Behind, it's behind but consistent (outdated as stated in the docs)? As I've said, our write load is low enough that Ahead/Behind does not happen too often and if the state is Connected, then it is not too far behind (besides what is left in the network buffers and/or not yet hit secondary disk due to protocol A)? Regards, Saso Slavicic -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.linbit.com/pipermail/drbd-user/attachments/20150127/66a6d0de/attachment.htm>