Hello, I have a problem with a knot slave instance, which does not refresh anymore. When doing a retransfer, I have a "no usable master" error, with this error log: oct. 21 11:02:16 corrin knotd[32453]: info: [geekwu.org.] control, received command 'zone-retransfer' oct. 21 11:02:16 corrin knotd[32453]: 2019-10-21T11:02:16 info: [geekwu.org.] control, received command 'zone-retransfer' oct. 21 11:02:17 corrin knotd[32453]: info: [geekwu.org.] AXFR, incoming, remote 10.42.42.6@53, starting oct. 21 11:02:17 corrin knotd[32453]: 2019-10-21T11:02:17 info: [geekwu.org.] AXFR, incoming, remote 10.42.42.6@53, starting oct. 21 11:02:18 corrin knotd[32453]: debug: [geekwu.org.] refresh, remote master, address 10.42.42.6@53, failed (connection reset) oct. 21 11:02:18 corrin knotd[32453]: 2019-10-21T11:02:18 debug: [geekwu.org.] refresh, remote master, address 10.42.42.6@53, failed (connection reset) oct. 21 11:02:18 corrin knotd[32453]: 2019-10-21T11:02:18 warning: [geekwu.org.] refresh, remote master not usable oct. 21 11:02:18 corrin knotd[32453]: 2019-10-21T11:02:18 error: [geekwu.org.] refresh, failed (no usable master) oct. 21 11:02:18 corrin knotd[32453]: warning: [geekwu.org.] refresh, remote master not usable oct. 21 11:02:18 corrin knotd[32453]: error: [geekwu.org.] refresh, failed (no usable master) On master, I have this error log: oct. 21 11:02:16 arrakeen knotd[20289]: info: [geekwu.org.] AXFR, outgoing, remote 10.120.42.42@33435, started, serial 2019100808 oct. 21 11:02:16 arrakeen knotd[20289]: 2019-10-21T11:02:16 info: [geekwu.org.] AXFR, outgoing, remote 10.120.42.42@33435, started, serial 2019100808 oct. 21 11:02:16 arrakeen knotd[20289]: debug: TCP, send, address 10.120.42.42@33435 (connection timeout) oct. 21 11:02:16 arrakeen knotd[20289]: 2019-10-21T11:02:16 debug: TCP, send, address 10.120.42.42@33435 (connection timeout) tcpdump shows a TCP connection from 10.120.42.42 to 10.42.42.6, which is closed with a TCP FIN by 10.42.42.6 after 1.2s Do you have hints about that ? Thanks,

Le lundi 21 octobre 2019 à 11:25 +0200, Daniel Salzman a écrit : You're right, thanks ! By the way, documentation says "Set to 0 for infinity.", but when I set 0, the connection is closed immediately. Setting 10000 solved my problem.

Regards,