Note: "permalinks" may not be as permanent as we would like,
direct links of old sources may well be a few messages off.
Hi all: I have a drbd 8.3.13 compiled on centos 6.2(kernel 2.6.32-220.el6.x86_64) I tried the Ahead/Behind mode by setting "on-congestion pull-ahead". Thought I don't exactly understand what does the 'congestion-fill' parameter mean, by changing it's value, I noticed something amazing. The manual said the optional values is 0 to 10GiBytes; in my case, I tried 256K, 4M, 8M, 32M, and 256M, and keep writing data on the drbd device: with value of 32M or 256M, it works as usual (cs:Connected); with value lesser than / equal to 8M, Connection state changed from 'Connected' to 'Ahead/Behind'. However, after I stop I/O writing, the value of 'oos' keep increasing for a short while (that's reasonable) and stopped at certain value. after that moment, the replication seems to be stopped, the 'oos' never changed again with 'cs:Ahead/Behind' & 'ds:Outdated/UpToDate' , I'm sured that there's no more I/O on that device. Unless I disconnect & reconnect it, it would do resync again. So my questions are: 1. what dows 'congestion-fill' exactly mean? 2. is there something wrong with the above situation?