HAST primary role hang when copying data

Solène Rapenne solene at bsd.zplay.eu
Mon Jan 25 18:03:45 UTC 2016


I am trying to use HAST between 2 FreeBSD-10.2 servers which are in 2 
differents DC with a 100/100 mbit/s network and 14ms of ping between 

I used both (not at the same time) OpenVPN and a SSH Tunnel to transport 
the data for security.

In both tries, hastctl tells me that the sync is complete, I can mount 
it on one one, then I start to cp some files inside and then the primary 
system hang after a few seconds when I do something disk related, only a 
hard reboot can fix this. I can reproduce it anytime on both nodes. When 
starting to write on the primary, the second node lose synchronization 
and the primary hang like if it has a nfs mounted on something 
disconnected (if you know that case).

I tried some hast.conf "tweaks" from the man to try to make it more "vpn 
friendly" without success

resource essai {
     replication async
     compression lzf
     on BBB {
         local /home/shared
         metaflush off
     on AAA {
         local /home/shared
         metaflush off

/home/shared is a file made with dd with zeroes, maybe this is the 
problem ?

Am I doing something wrong ?

Kind regards

More information about the freebsd-questions mailing list