One masternode becomes inactive after other restarted (`mns teeter-totter`)

GilAlexander

Member
Jun 26, 2014
84
23
48
What I have:
Two masternodes (lets A and B) and configured start-many to start them. Local: Windows 7 x64, wallet.dat with keys for A and B, masternode.conf, remote: ubuntu x86_64, dash.conf with mnprivkey, ufw with opened 9999, no iptables, dashwhale v4 script for monitoring. Last versions of all dash clients. All the clients reinstalled from the scratch.

The sequence of events that cause the problem:
1. I started both of masternodes using "start-all", they all are active and got paid.
2. For some external fair reason (I've upgraded the server) B became inactive (and had been inactive for a while). A is still active.
3. I opened my local wallet and did "walletpassphrase *** 600", "masternode start-disabled". B was started and became active.
4. After an hour A became inactive (EXPIRED, then vanished from the list). I took a look at dashninja.pl and had noticed that A's last seen = 1h15m before it finally vanished. That's why it expired. I looked at A's daemon and it was fine - no crashes, no stuck, fully synced and >8 connections.
5. I opened my local wallet and did "masternode start-disabled" again. Now A was started and became active.
6. One can guess what's next. A is good now and B have "Last Seen" values more and more. B's daemon is also fine, I've even restarted it (when B is active but Last Seen=45m) and that didn't help.
There are two future evolutions:
a) Both "Last seen" will reset by themselves and all is good for a long time; it may take some more "masternode start-disabled" commands (that will start A and B alternately). "Last seen" will never be more than 15m.
b) I'll do "masternode start-all" (or wait for both mns become inactive, then "masternode start-all") and they will work properly.

I've used "masternode start" more than a year and never had such problems. I've moved to start-many a week ago and have already issued that two times (so it's a repeatable pattern, for me at least). What am I doing wrong? I wonder how do the "last seen" can grow when the daemon is fine and responsive and why the A's start spoils B and vice versa?

P.S. I have another observation. When I used "masternode start", the list entry about my masternode appeared at once on all of my servers - and very quickly on external services like dashninja.pl and others. When I use "masternode start-all", it takes some more time (<5 min) to appear in the list even on the mn's hot wallet.
 

GilAlexander

Member
Jun 26, 2014
84
23
48
The problem is: I used the same masternodeprivkey on several masternodes.

I experienced the same problem as described after 3 months when I hit "start-disabled" once more after one mn had crashed. The interesting thing that mns can work properly even with the same masternodeprivkey - and they worked for 3 months. The problem occures when you start them or trying to do smth with one of them.

Foolishly - I checked so many things before I figured the problem out - I really thougt that masternodeprivkey must be identical for every mn in "start-many" configuration. :D
 
Last edited by a moderator:
  • Like
Reactions: TaoOfSatoshi

splawik21

Grizzled Member
Dash Core Team
Moderator
Foundation Member
Dash Support Group
Apr 8, 2014
1,917
1,274
1,283
Great :)
The best lessons you learn are on your own mistakes ;) but it is always better to learn from success of others and ask for help.